Ebook About Biology Research AAA

MED I CINE W I T H O U T MED S
Ho_Meds_int_3pgs.indd 1 14/06/23 12:45 PM

This page intentionally left blank
349-99188_Rothfels_ch01_3P.indd 6
Medicine
without Meds
•
Transforming
Patient Care with
Digital Therapies
•
DE AN HO
YOANN SAPANEL
AG ATA BL ASIAK
JOHNS HOPKINS UNIVERSIT Y PRESS | Baltimore

2023 Johns Hopkins University Press
©
All rights reserved. Published 2023
Printed in the United States of America on acid-f ree paper
9 8 7 6 5 4 3 2 1
Johns Hopkins University Press

2715 North Charles Street
Baltimore, Maryland 21218
www.press.jhu.edu
Library of Congress Cataloging-in-Publication Data is available.
ISBN 978-1-4214-4703-2 (hardcover)

ISBN 978-1-4214-4704-9 (ebook)
A catalog record for this book is available from the British Library.
Special discounts are available for bulk purchases of this book. For more information,
please contact Special Sales at specialsales@jh.edu.

Contents
Foreword, by D. A. Wallach vii

Preface xi
Introduction. The Context 1
PART I. WHAT ARE DIGITAL THER APEUTICS? 19

1. Digital Health Solutions versus
Digital Therapeutics 21
PART II. THE RIGHT TOOL FOR THE RIGHT PROBL EM 39

2. Addressing Unmet Stakeholder Needs 43
3. Addressing Unmet Clinical Needs 77
4. Ensuring Business Viability Early 91
5. Test, Learn, and Pivot 101
PART III. FROM PIL OT TO PRODUCTION 113

6. Data Sourcing and Management 117
7. Clinical Validation: Unique Challenges and
Novel Approaches 128
8. Factors That Influence the Implementation of
New Technologies 154
9. Recommendations for DTx Implementation 162
10. Supporting Adoption of DTx 173

PART IV. PATHS TO COMMERCIALIZ ATION 191
11. Channel Partnerships: The Ultimate Commercialization
Strategy 197
12. The Price Is Right: Value in the Eye of the Beholder 231
Conclusion. The F uture of DTx 249

With contributions by Dr. Eddie Martucci
Acknowledgments 257
Appendix A. AI and Its Potential Use in DTx: A Brief Overview of Frequently
Asked Questions 259
Appendix B. How to Translate Evidence-Based Theories to the Design of Digitally
Delivered M
ental Health Interventions 270
Bibliography 277
Index 305
vi Contents

Foreword
It Is Time for Health Care to Embrace Patients
D. A. Wallach
*an industry-defining company:
a business or organization that has had a profound
and lasting impact on its respective industry.
be poised to = be ready to
D. A. Wallach is a venture capitalist and health care innovator focused
on technologies poised to reinvent medicine. Industry-defining
companies backed by the funds he cofounded—Time BioVentures
and Inevitable Ventures—have included Beam, Glympse, Doctor on
Demand, Devoted Health, and Neuralink. Wallach is an acclaimed
recording artist and active in the nonprofit space focused on
preventive medicine and patient support.
Through the Lens of a Consumer and Tech Investor

Early-stage tech is fascinating. As an investor, I follow related devel-
opments on both micro and macro levels: advising startups at the
academic stage while observing global trends being shaped by (and
shaping) consumer demands. While I started my investing journey in
technology with companies including Spotify, SpaceX, and Ripple, over
time I developed a keen interest in health care. Carrying over my under-
standing of the dynamics of innovation to this sector, I am impressed by
how much room for improvement t here is in terms of consumer experi-
ence. Health care is one of the last frontiers where this shift in focus is
still just beginning.
Coming into the health care field from the perspective of an inves-
tor “trained” in consumer technology, I was, and still am, shocked by
the l imited role consumers—patients—play in driving the demand for
new solutions and high-quality user experiences. Instead, innovation
in this field is commonly stewarded by payer demands, even though
managed
in the health care systems of developed markets payers are typically
not the patients, but private or public insurers. Their motivations and
vii

objectives, as well as t hose of other stakeholders, may not be aligned
with the needs and desires of patients—the population that the sys-
tem should aim to serve.
motive Beyond this misalignment of stakeholders, t here is a general lack
of impetus for change on the part of gatekeepers with an interest in
maintaining the status quo. The traditional incentive system behind
technology adoption—staying competitive—is inoperative in health
care, which is dominated by decision-making processes between pay-
ers (insurers) and providers (hospitals and clinics) that are not well un-
derstood. For potential investors interested in the sector, this opacity
intransparency
translates into a reluctance to take on investment risk without hav-
ing a clear vision of the systems’ workings and reliable d
rivers of f uture
returns. In short, investors are wisely skeptical of the potential for in-
novation alone to produce successful, rapidly scaling companies. Yet,
without private funding much of the innovation potential simmering
in early-stage health technology, startups cannot be realized, result-
ing in an abundance of great technologies that languish at the proof-
of-concept or pilot stage.
But there are reasons to be hopeful. The COVID-19 pandemic that
put the world on pause and pushed much of health care delivery outside
the hospital and into people’s homes may well turn out to be an inflec-
tion point that shifts the balance of power. Many of the digital solu-
tions for remote patient monitoring, adherence to treatment, and dis-
ease self-management that emerged in the aftermath of this pandemic
may be h ere to stay, and they hold the promise of finally empowering
patients with the decision-making power they never quite had before.
Building Ventures to Reinvent Health Care

Without a doubt, the increased openness to remote patient care solu-
tions paired with unprecedented advances in digital technology will
serve as catalysts for health care innovation. Putting patients at center
stage requires a technological engine that personalizes care with treat-
ments optimized to each patient according to their genetic and physio-
logical profile, lifestyle, and needs. This space is ripe for new technolo-
viii Foreword

*bud: to produce buds
*granular: made of, or seeming like, granules（顆粒）
gies that can monitor, diagnose, and treat an affected population
without losing the granular level of focus on each individual patient.
The time is right for budding entrepreneurs in this space to get in-
vestors’ attention and the support they need to turn their ideas into
v iable, scalable products. A successful digital health endeavor is a con-
vergence of substance and sales. Substance means technology, team,
operation, execution, business strategy, and timing. Sales means the
entrepreneur’s ability to convince investors, employees, buyers, and
end users of the value and impact of their solution by sharing a vision
that is too good to miss. Being successful at balancing these elements
is a unique predisposition and a rare one to come across.
To build a venture, entrepreneurs must have strong self-awareness
and team-building skills. They need to understand their shortcomings
and be willing and able to address them by surrounding themselves
with the right people. Thus, a founder who is a technical wizard would
be wise to team up with an experienced business strategist and an ef-
fective communicator; one who is a physician by training may want to
join forces with a user-experience researcher, a software developer, a
marketer, or so on. The most innovative companies are those formed
by groups of p eople with complementary skills.
*predisposition: the state of being likely to behave
in a particular way or to suffer from a particular disease
What’s Next?
Digital health has the potential to reinvent health care. Wider stake-
holder acceptance notwithstanding, it is within the reach of digital
health entrepreneurs to tip the scales and transform piecemeal digi-
tal health inventions into large-scale health care innovation that em-
braces patients. It takes audacity to try and wisdom to know what the
process entails. But while audacity may come naturally to an entrepre-
neur, where can one find the wisdom? Reading this book is a good
place to start. *notwithstanding: despite the fact or thing mentioned
*audacity = bravery
Foreword ix

Preface
Traditional health care has a new ally. Patients with sleep disorders,
back pain, and diabetes are now being prescribed app-based treat-
ment in place of a drug. Algorithms are helping cancer patients to
manage their symptoms. Video games are improving the attention
span of children diagnosed with ADHD.
A new class of medicine, called digital therapeutics (DTx), is gain-
ing traction and transforming the way patients engage with the health
care system. Yet, as a patient, you may have never heard of this bur-
geoning field.
In this book, we speak to innovation and business leaders from all
corners of the health care universe who are involved in designing and
building tomorrow’s DTx. They include clinicians, researchers, engi-
neers, patients, startup founders, and corporate executives. We share
their insights on how DTx can deliver value beyond the technology, ad-
dress the challenges of implementation in existing health care mod-
els, and propose a way forward by revolutionizing care delivery.
Through a learning lens, we also take a look at DTx innovations to date
and analyze the key factors for their successes as well as their failures.
It is time to realize the potential of DTx, and we are bringing all
stakeholders along on this journey that w ill ultimately bring re
imagined health care to all.
xi

MED I CINE W I T H O U T MED S

• Introduction
The Context
You know that discomfort you feel in your lower back e very once in a
while? Or that sharp pain shooting down your leg when you stand up
too quickly? You have probably experienced low back pain at some
point in your life, which is similar to over 80% of the population.1 How
have you dealt with it? For 56-year-old Simon, a “slight discomfort”
(as he described it) did not stop him from his daily activities at first,
so he didn’t really pay attention to it. Over the years, though, the pain
became more present, persistent, and sharp—evolving from a dull
ache to chronic pain. Eventually he even had to take time off from
work. Walking the dog became difficult. Playing with his kids was a
struggle. Lifting grocery bags was impossible. painkiller
Despite remarkable advances in medicine, a common health prob
lem such as low back pain is still mainly treated with analgesics. Sadly,
they are effective in relieving the symptoms but not the underlying
cause, which often remains unknown and therefore unaddressed. To
complicate m atters, Simon doesn’t remember when the pain started
or what triggered it. He simply knows, as his doctor had repeatedly told
him, that he should move more, lose some weight, and lead a more active
lifestyle. But do we all always follow our doctor’s advice?
1

Over the years, Simon has had many consultations—not only with
his family doctor but also with chiropractors, physiotherapists, and
pain therapists—in search of a remedy. Physical therapy, rehabilita-
tion, spinal manipulation, you name it: Simon has tried them all and
spent an inordinate amount of time and money doing so. Fortunately,
Simon’s doctor was able to suggest a new approach—a mobile app for
chronic pain relief. Hopeful that it would help him take control of his
pain, Simon did not hesitate, especially since the solution promised to
take only 10 minutes a day.
A medical and technological novelty for Simon, the app uses virtual
reality supported by the latest scientific evidence in digital training,
pain education, and psyc hology to deliver a behavioral intervention
that may work as well as, and maybe even better than, medication.
Over the course of a few weeks, the app “nudged” Simon to take a com-
prehensive approach to managing his pain through:
• Exercise: tailored to alleviate his pain and keep his body active;
• Relaxation: mindfulness, breathing, and progressive muscle
relaxation techniques; and
• Knowledge: general pain and back pain–specific education
modules to control the perception of pain.
Guided by the app, Simon received this regimen that adjusted itself to
match his progress. For instance, when the app detected his motiva-
tion waning, a dedicated coach “intervened” in an attempt to reener-
gize him through interactive question and answer sessions.
After three weeks of daily use, Simon began to notice improvements
as he gradually learned how to take control of his pain. This motivated
him to finally start making adjustments to his sedentary lifestyle and
in four months, Simon was pain-f ree for the first time in 20 years.
Today, he no longer needs his pain medication—what a relief! If life is
better without painkillers, it is much better without any pain at all.
Low back pain is the leading cause of activity limitation and absen-
teeism from work, resulting in significant medical burden and eco-
nomic cost.2 Consequently, it has become a major global public health
2 MEDICINE WITHOUT MEDS

problem.3 As the world population ages, the prevalence of low back pain
is projected to increase substantially due to spinal deterioration of
older p eople in ever greater numbers.
Fortunately, health-focused technology, such as the mobile app that
helped Simon gain control over his back pain, offers hope to the 540
million individuals estimated to be experiencing low back pain at any
one time.4 These types of apps represent a new class of medicine called
digital therapeutics (DTx). Like drugs, DTx have to fulfill stringent
requirements that demonstrate their safety and efficacy through rig-
orous clinical trials before they can be prescribed (we w ill discuss this
at length l ater in the book). But most importantly, beyond back pain,
DTx are clinically useful for a broad spectrum of behavioral, mental,
and physical diseases and disorders. As such, they are regulated as
“software as a medical device.”
While we chose to share Simon’s story of dealing with low back pain,
we could have also shared the stories of Sarah, Elena, or Jin and how
they are dealing with high blood pressure, high blood glucose, hyper-
tension, excess weight, or early signs of dementia. Globally, one in three
adults live with more than one chronic condition, whereas in the United
States alone up to 60% of Americans have one or more preventable
chronic diseases.5 Examples include:
• Cardiovascular diseases: the leading cause of death globally,

taking an estimated 17.9 million lives each year.6
• Diabetes: the world’s fastest growing chronic condition and
already one of the most costly and burdensome chronic dis-
eases.7 One in 11 adults has diabetes and one in two adults with
diabetes is undiagnosed.8
• Cancer: one in two people w ill be diagnosed with cancer in their
lifetime.9
• Dementia: someone in the world develops dementia every three
seconds, adding up to nearly 10 million new cases every year.10
The list goes on b

ecause society is aging. We are living longer but not
necessarily healthier.
Introduction 3

*twofold
as noun.
1. twice as big or as much 2. having two parts
as adverb.
2. by two times e.g. The sport saw its popularity increase twofold
The consequences of an intensifying chronic disease epidemic are

twofold. Firstly, chronic diseases represent a significant burden to
those who live with them, as we observed with Simon. Most of us know
someone who suffers from diabetes, cardiovascular disease, cancer, or
a mental health disorder. The daily experiences of t hese patients, fam-
ilies, and caregivers who are living and coping with t hese conditions
and their side effects should not be neglected. Secondly, from a sys-
tems perspective, managing t hose diseases puts an ever-growing pres-
sure on total health care expenditures, which in turn makes health
care less accessible. In the United States, 90% of the $3.5 trillion annual
health care spending is dedicated to chronic and mental health diseases,
and this expenditure is projected to continue to outpace gross domes-
tic product growth over the next 15 years.11 And this upward cost
spiral is not unique to the United States.
This unprecedented rise in chronic and mental health diseases, cou-
pled with the continually growing costs that accompany them, require
a fundamentally new approach to how we conceptualize and deliver
care. What role could DTx play in transforming our health care systems
to match the size of this challenge?
Existing Solutions in Health Care Are Not Adopted

or Scalable (Enough)
New health care delivery models are emerging. Compared to estab-
lished pathways, they are more data-driven and rely to a greater de-
gree on digital biomarkers, digital diagnostics, and DTx to measure,
diagnose, monitor, and manage patients’ health. As such, some of them
are providing an entry for DTx in care circuits in recognition of the
potential to fulfill unmet needs and enhance patient-centered care.
Early adopters of new medical technologies believe in the potential
of DTx to disrupt traditional and often inefficient health care business
models through delivering better clinical outcomes at lower cost while
providing a superior patient experience. In fact, DTx are said to be one
of the most promising avenues for improving health care provision on

a global scale, so much so that some even see them as poised to revo-
lutionize medicine. In our view, they w ill play an important role in the
transformation of our health care systems from “sick care” to “health
care” and “people care” that’s outcome-d riven, evidence-based, and
experience-focused.
DTx are certainly not g oing to be cure-alls for the ailments afflict-
ing global health care systems. Although on one hand they offer great
promise, Simon’s story of using a DTx to manage his pain symptoms,
on the other hand, remains anecdotal. DTx are relatively new to the
place where you store your weapons
health care arsenal and still in the early days of commercialization.
What this means is that patients and caregivers are more likely to have
never heard of such solutions or how they can be helpful. And physi-
cians already struggle with myriad mobile applications, each promis-
countless
ing to deliver better health outcomes to their patients, so they may be
skeptical to try “yet another app.” And who is going to pay for all of these
tools anyway?
T hese considerations give rise to important questions:
• How w ill DTx be implemented?

• How can DTx adoption be expanded?
• Can we harness DTx to broadly decentralize health care?
Technology Alone Cannot Transform Health Care

Taking a technological innovation all the way to the finish line—the
clinic or home—involves a wide range of disciplines and expertise,
from behavioral science, design thinking, and software engineering to
user experience research, gamification, and health care economics. It
also depends on inventors and developers being part of a well-oiled
ecosystem whose discrete parts are willing to collaborate, often across
value chain segments and even country borders. Thus, realized inno-
vation lies not only in the technology itself, but also in how that
technology is tested and commercialized, how patients and physi-
cians are engaged, and how all other stakeholders are partnered. This
Introduction 5

Missed Opportunities Will Be Costly
Unlike many customer-facing industries (e.g., hospitality, retail, and banking), which
have all been transformed by the adoption of new technologies, health care has
been slow to open itself to innovation. One only needs to look at the current state of
digitization in the health care industry; patchy solutions notwithstanding, it remains
an aspiration. To cite one statistic that speaks to this reality, over 12% of Germany’s
total health care costs (34 billion euros, or US$40 billion), could have been saved in
2018 if the German health care system had been fully digitized.12 Where would the
largest savings and impact have come from? The switch to paperless data via
electronic health records (EHRs), which w
ere in fact developed in the 1980s.
EHR systems, like many health technologies (including DTx), belong to the
category of innovations presenting important potential benefits. However, their
mainstreaming in routine clinical settings remains difficult. Despite billions in
incentives spent on expanding their adoption and use globally, the central question
about EHR interoperability across all settings of care has yet to be resolved. Further,
we still know too little about how EHRs are being used to improve outcomes while
drawing on fewer resources. In fact, one study found that it is the opposite: for every
hour of clinical work, physicians spent two hours on clerical or EHR-related tasks.13
related to paper work
The EHR example underscores that technology alone, even if successfully
operationalized and implemented, does not guarantee wide market adoption. For
this reason, DTx entrepreneurs should make sure their innovations effectively
bridge the gap between development and commercialization. This book contains
actionable strategies on how to achieve this outcome.
interactional element is what opens doors and pushes new bound

aries in terms of technology implementation that improves lives for
patients, caregivers, and communities across the globe while also em-
powering health care professionals.
The Institute for Digital Medicine (WisDM) of the Yong Loo Lin
School of Medicine at the National University of Singapore was estab-
lished with the mission to harness leading-edge technology and forge
mutually supportive partnerships between stakeholders involved in

its development and commercialization, in order to drive practice-
changing medicine that achieves scalable patient impact. In this journey
we are guided by a s imple but ambitious goal: to develop a framework
for implementing and scaling novel pharmacological and digital inter-
ventional technologies into yet-to-come models of care.
Granted, the transformation of our health care system has already
been set in motion by advances such as artificial intelligence (AI), tele-
health, DNA sequencing, personalized medicine, and DTx. We are al-
ready transitioning toward a health care system that is more proactive
than reactive, more digital than analog, and more data-driven and per-
sonalized than one-size-fits-all; a system that will also be more patient-
or user-centric, focused on restoring health to individuals rather than
on treating body systems and organs in isolation.
We must recognize, however, that technology alone cannot trans-
form health care. Numerous real-world examples have demonstrated
that substantial gaps exist between major capital raises and clinical
uptake and reimbursement.14 Transforming health care is indeed about
more than introducing new technology; it is also about improving
patient outcomes and doing so at a lower cost than under the status quo.
Patients are not demanding new technologies just for the sake of
it. They r eally just want to feel and get better. They need help to treat
and manage their conditions, but the how—whether by taking a pill,
using a mobile app, or benefiting from an algorithm—is less impor
tant to them than the what, which is the final outcome. Technology
should seamlessly integrate into their day-to-day lives, to deliver both
a seamless experience and improved quality of life.
Physicians also d on’t need technology just for the sake of it; what
they need is better support to manage their patients in a more individ-
ualized way. This can potentially be achieved via access to more granu-
lar information that lets them continuously optimize patient care. “It is
not b ecause you have an algorithm that you have a commercial-ready
product,” says Dr. Carolyn Lam, cofounder of the Singaporean medtech
company Us2.ai, which uses machine learning to automate the analysis
of echocardiograms. “We should not underestimate the amount of work
Introduction 7

that it takes to integrate with existing EHRs, fix data silos, design the
right interface, protect patient data, get all the regulatory approvals,
et cetera. It is equally demanding, if not more demanding than it is to
develop a new technology,” she tells us.
In a nutshell, technology is an enabler but not a solution in and of
itself. Getting the technology right is just the beginning of a long
and sinuous journey for DTx, often marked by continuous struggles
if not outright failure. In reality, a global community of innovators
striving to improve patient outcomes are facing the same obstacles.
Ultimately, the right innovation blueprint w ill realize the full potential
of DTx.
Who Is This Book For?

In this book, we speak to innovation and business leaders from all cor-
ners of the health care universe who are involved in designing and
building tomorrow’s DTx. They include clinicians, researchers, engi-
neers, patients, startup founders and corporate executives—all driven
to ultimately improve patient health outcomes.
DTx represent a convergence point of three diff erent worlds: health
care, technology, and business. Developing and commercializing a DTx
solution requires a deep understanding of each of t hese diff erent uni-
verses and their vernacular. Which language do you speak?
• If you “speak” health care, you may be a clinician, nurse, serv ice
lead, or hospital administrator. As such, you understand patients’
needs and frustrations, as well as the infrastructural challenges
of providing quality patient care under the status quo. You
could envision a new DTx solution to benefit your patients (and
potentially your practice). The next steps you w ill need to
address is how you develop a DTx technology that embodies
your idea while also ensuring viability in practice.
• If you “speak” technology, you may be a software developer,
engineer, data specialist, or chief technology officer. As such,

you approach the development of DTx in the context of techni-
cal performance, including things like processing speed, software
architecture, how much computing power it consumes, and
maybe even cybersecurity. You may also be thinking about how
to properly determine its effectiveness. In order to be useful,
however, a health care technology must be successfully taken
out of development, established with evidence, and ultimately
implemented.
• If you “speak” business, you may be an entrepreneur, corporate
executive, C-suite leader, or an investor. You likely work in a field
such as marketing, advertising, finance, sales, or management.
When you think about DTx, you understand the need to size up
an opportunity, to evaluate its potential impact on the end users,
and to anticipate challenges involved in implementing a new
solution in the current health care ecosystem. Successfully
addressing financial return and scalability in the DTx journey
ahead w ill be essential.
Realizing Medicine 4.0

This book serves as a blueprint to bring digital therapy to fruition and
as a gateway to inspiring unprecedented broader impact on the f uture
of medicine. We therefore aim to offer an end-to-end roadmap for DTx
solution scalability that w ill ultimately help DTx innovators and en-
trepreneurs deliver impact at scale. Our perspectives and insights come
from both academia and the technology industry, including stories and
learnings that address the practical realities of DTx development and
implementation. We also look within the health care industry for guid-
ance on successful implementation in the clinical setting. Lastly, we
take a look at existing as well as aspiring DTx innovations and learn
from both their successes and failures.
Along with the frameworks, tools, and insights this book contains,
it w
ill prepare you for the inevitable twists and pivots your organization
may experience as it gears up for entering the DTx space. It may not
Introduction 9

provide all the answers, but we hope it w ill give you a solid founda-
tion for derisking your journey ahead.
A fter introducing what DTx are and what’s unique about them
(part I), we explore the main challenges and opportunities for each
key development phase of DTx organizations, from the inception of
the organization and exploration of potential concepts (part II), to
validation of the DTx solution in routine clinical care (part III), and
finally the go-to-market strategy and how to deliver impact at scale
(part IV).
Part I: What Are Digital Therapeutics? As a new technology, it may be
unclear what DTx are and what makes them unique. DTx should not
be confused with wellness apps. The distinction between the two is
clearly defined by the level of clinical evidence, real-world outcomes,
and regulatory oversight involved in their development and evalua-
tion. A fter reading this first chapter, innovators and business leaders
curious about the f uture of medicine w ill understand the critically
important role of DTx toward changing the way that medicine is
practiced.
Part II: The Right Tool for the Right Problem. DTx are a versatile class
of medical tools that offer a range of benefits. What kinds of problems
are they best positioned to address? In chapters 2–5, we invite readers
to develop their thinking about DTx across three dimensions: clinical
needs, stakeholders, and venture viability.
Part III: From Pilot to Production. Exploring the real-world potential of
DTx is about testing how well they operate in routine clinical practice.
During this stage, DTx transition from invention to innovation if their
creators are able to demonstrate they represent a solution in day-to-day
clinical care. This phase is a litmus test before scale-up and chap-
ters 6–10 guide DTx innovators through the complexities of implemen-
tation and adoption of DTx to arrive at the right solution-market fit.
Part IV: Paths to Commercialization. Having a good, tested idea and a
minimally v iable product do not guarantee commercial success. How
do you reach the end users (patients)? How much should you charge

for your digital solution? In the final two chapters we explore the
importance of having the right strategic partner to distribute and
commercialize your DTx and discuss monetization strategies.
Introducing the Key Contributors

This book goes beyond our team’s experience in developing and imple-
menting new technologies in health care. It is the result of a collabora-
tive approach that also leverages the experience and know-how of
many external distinguished experts in fields such as AI and imple-
mentation science. We are privileged to share their thoughts with our
readers.
Listed in order of appearance in the book:
Dr. Jessica SHULL

Head of Digital Therapeutics at Vicore Pharma, Sweden
What Are Digital Therapeutics?, page 19
Dr. Jessica Shull is Head of Digital Therapeutics at Vicore Pharma
in Sweden. Previously, she led the European Policy work for the Digital
Therapeutics Alliance and managed clinical projects for respiratory
disease at Bellvitge Hospital in Barcelona.
With 20 years’ experience in digital health—from developing vir-
tual surgery devices to integrating digital frameworks at the World
Health Organization and at national level—she is now focused on
developing digital therapeutics products.
Dr. Ester YEOH

Senior consultant endocrinologist at Admiralty Medical Centre and
Khoo Teck Puat Hospital, Singapore
Addressing Unmet Stakeholder Needs, page 43
Dr. Ester Yeoh is trained in endocrinology and internal medi-
cine, with subspecialty interest in type 1 diabetes and hypoglycemia.
She leads the Flexible Insulin Therapy team at the Admiralty Medical
Centre’s Diabetes Clinic, which cares for people with type 1 diabetes.
Introduction 11

The team comprises a dietician, diabetes nurse educators, and a
health psychologist.
Dr. Yeoh firmly believes in patient education and in empowering her
patients with the knowledge and tools to improve diabetes self-
management. She has special interests in patient education, m ental
health in diabetes, and diabetes technologies such as insulin pumps,
glucose sensors, and other digital health solutions.
Dr. Raj KUMAR

CEO of UCSI Blue Ocean Strategy Consulting, Malaysia
Dr. Raj Kumar is currently the CEO of UCSI Blue Ocean Strategy
Consulting, a consulting firm that specializes in strategic value in-
novation. He is a certified Blue Ocean Strategy expert whose con-
sulting insights draw on his vast experience, ranging from setting
up business operations to strategic planning, innovation, and project
execution.
Dr. Kumar has helped implement Blue Ocean Strategy for interna-
tional firms, billion-dollar conglomerates, and government agencies.
He has also conducted programs on Blue Ocean Strategy across North
America, the Middle East, Africa, and the Asia Pacific region, interact-
ing with participants from over 25 countries.
Fabien SOUILLARD
Type 1 diabetes patient, France
Fabien Souillard was diagnosed with type 1 diabetes when he was
13 years old. Ten years l ater, a fter countless consultations with endo-
crinologists and other physicians, he still struggles with daily finger
pricks, carbohydrate counting, and trying his best to have the sem-
blance of a “normal” life. He often needs to visit hospitals and man-
age his hemoglobin A1c, stomach cramps, and other complications that
arise from his condition.

Dr. Robyn MILDON
Founding Executive Director of the Centre for Evidence and
Implementation, Australia
Recommendations for DTx Implementation, page 162
Dr. Robyn Mildon is an internationally recognized authority in the
fields of implementation science, evidence synthesis, and knowledge
translation, as well as program and policy evaluations in health, edu-
cation, and h uman serv ices. She is an adjunct associate professor at
Monash University, a visiting professor at the Yong Loo Lin School of
Medicine, codirector of the recently established the Centre for Be-
havioural and Implementation Science Interventions at the National
University of Singapore, and the inaugural cochair of the Knowledge
Translation and Implementation Group with the Campbell Collabora-
tion, an international systematic review group.
Over her career, Dr. Mildon has helped to advance the implemen-
tation of better evidence in policy and practice settings and to im-
prove the quality and effectiveness of health, education, and human
ser v ices.
Dr. Kee Yuan NGIAM

Group Chief Technology Officer, National University Health System,
Singapore; Deputy Chief Medical Informatics Officer, National
University Hospital, Singapore
Recommendations for DTx Implementation, “What Health Care System
Managers Expect,” page 162
Dr. Kee Yuan Ngiam is the Group Chief Technology Officer of the
National University Health System (NUHS) Singapore, overseeing
the technology deployment in the Western health care cluster. In this
role he assists the system’s chief executive to implement new technol-
ogies throughout the NUHS and serves as chief advisor to the NUHS
Centre for Innovation in Healthcare. Dr. Ngiam is concurrently the
Deputy Chief Medical Informatics Officer of NUHS, with a special
focus on artificial intelligence research and implementation in health
care. In his capacity as assistant professor of surgery at the Yong Loo
Introduction 13

Lin School of Medicine at the National University of Singapore, he also
engages in research into endocrine and metabolic surgery.
Dr. Ngiam led the operationalization of the DISCOVERY AI plat-
form—a translational research-to-production AI platform that is
the basis for operational validation of advanced deep-learning AI
models—by encouraging collaboration between computer scientists
and clinicians. Since its launch, the platform has promoted interdisci-
plinary collaborations, reducing the time to translation and acceler-
ating the pace of health care innovation.
Christopher HARDESTY
Partner, Pureland Venture, Singapore
Paths to Commercialization, page 191
Christopher Hardesty specializes in building early-stage medical in-
novations for scale-up in Asia and beyond. He has lived and worked in
more than 50 countries to design public-private health system finan-
cial schemes, including for the creation of pathways for adoption of
medical innovations. He is also a member of KPMG’s Global Health-
care & Life Sciences Centre of Excellence and engages in adjunct lec-
turing, research/writing, and startup/fund advisory so as to be a stew-
ard of innovation acceleration across the health care ecosystem.
Dr. Eddie MARTUCCI

Cofounder and CEO of Akili Interactive Labs, United States
uture of DTx, page 249
Conclusion. The F
Dr. Eddie Martucci is a cofounder and CEO of Akili Interactive Labs.
A fter earning his doctorate, he studied health care entrepreneurship
as a Kauffman Fellow through support from the Kauffman Founda-
tion, a nationally recognized leadership program for venture capital
ists and innovators. For the last 11 years, Dr. Martucci has worked
at Boston-based PureTech Health, a health care biotech company,
where he helped launch their digital health initiative and cofounded
two health-focused startups. In 2012, he cofounded and currently
serves as CEO for Akili Interactive Labs, a company recognized as pi-
oneering the new field of “digital medicine.”

Matt OON
Founder and CEO of Acceset, Singapore
Appendix B. How to Translate Evidence-Based Theories to the Design of
Digitally Delivered Mental Health Interventions, page 270
Matt Oon is the founder of Acceset (pronounced “asset”), a social
enterprise platform that aims to transform and normalize mental
health care in Singapore through providing anonymous text-based
emotional support for students via a digital letter-w riting portal.
Soon a fter founding the company, Oon received the 2018 Queen’s
Young Leaders Award in recognition of his commitment to revolution-
izing the way we perceive m ental health in the twenty-first c entury.
He was a member of the Wellcome Trust M ental Health Science Com-
munity, and he was the recipient of the Philip Yeo Initiative fellowship
and a member of the Generation T list, both initiatives recognizing
leaders of tomorrow who are shaping Asia’s f uture. His work ranges
from capacity building, such as the development of empathy skills
in students, to translational research work that involves implement-
ing the digital letter-w riting platform in schools to better support
students.
Finally, this book has also been shaped by input from 70 DTx and
digital health CEOs, entrepreneurs, innovators, clinicians, payers, and
industry experts across 15 countries in America, Asia, and Europe. We
acknowledge the contributions of:
Adrian Ang, AEvice, Singapore

Alexandre Tavares, Novartis, Singapore
Alexandria Marie Remus, N.1 Institute for Health, Singapore
Aline Noizet, Digital Health Connector, Spain
Amir Bozorgzadeh, Virtuleap, Portugal
Anthony Pannozzo, Frog Design, United States
Antonio Estrella, Taliossa, Singapore
Azran Osman-R ani, Naluri, Malaysia
Benjamin Belot, Kurma Partners, France
Introduction 15

Carolyn Lam, Us2.ai, Singapore
Charlotte Lee, Big Health, United Kingdom
Christopher Hardesty, Pureland Venture, Singapore
Christopher Lew, Rock Health, United States
Cole Sirucek, DocDoc, Singapore
Craig A. Delarge, The Digital Mental Health Project, United States
Danny Kim, Welt, K orea
Dorothea Koh, Bot MD, Singapore
Edward Booty, reach52, Singapore
Elena Mustatea, Bold Health, United Kingdom
Elizabeth Galbut, SoGal Ventures, United States
Eugene Hong, DBS, Singapore
Francois Cadiou, Healint, Singapore
Gabriel Sim, APACMed, Singapore
Geck Hong Yeo, N.1 Institute for Health, Singapore
Georgios Kaissis, Technical University of Munich, Germany
Girish Bommakanti, Access Health, India
Guillaume Bezie, Sibius, France
Hilary Thomas, KPMG, United Kingdom
Hsien-Hsien Lei, AMCHAM, Singapore
Jennifer Shannon, Cognoa, United States
Jeremy Lim, AMiLi, Singapore
Julian Sham, Amazon Web Serv ices, Singapore
Kitty Lee, Temus, Singapore
Maleena Suppiah Cavert, National University Health System,
Singapore
Markus Pratschke, Sandoz, Germany
Melissa Hudzik, Wilson Sonsini Goodrich & Rosati, United States
Miguel Rivera, Novartis, Singapore
Nana Bit-Avragim, iHospital InnoLab, Germany
Nang Duangnapa, Mercer, Singapore
Nawal Roy, Holmusk, Singapore
Neil Narale, Mercer, Singapore
Nicholas Brocklebank, Global Legal Solutions Group, Singapore

Owen McCarthy, MedRhythms, United States
Peta Latimer, Mercer, Singapore
Peter Hawkes, Johnson & Johnson, Singapore
Pierre A. Morgon, MRGN Advisors, Switzerland
Pieu RoyChoudhury, Medtronic, Singapore
Pin Kwok, Savonix, Singapore
Pocket Sun, SoGal Ventures, China
Renato De Giorgi, Axenya, Argentina
Reza Shokri, National University of Singapore
Rohit Bhatia, Cerebral, United States
Sajid Rahman, Digital Healthcare Solutions, Bangladesh
Sean G. Kang, Welt, Korea
Siva Nadarajah, JOGO Health, United States
Smrithi Vijayakumar, N.1 Institute for Health, Singapore
Snehal Patel, MyDoc, Singapore
Sue Anne Toh, NOVI Health, Singapore
Sundeep Lal, BioConnexUS, United States
Sven Jungmann, Founderslane, Germany
Theodore Wonpeum Kee, N.1 Institute for Health, Singapore
Timothy Johns, Department for International Trade, China
Valeria Burrone, MEng, Italy
Viroshini Hari Krishnan, AXA, Singapore
V Vien Lee, N.1 Institute for Health, Singapore
Wai Chiong Loke, Ministry of Health Office for Healthcare
Transformation, Singapore
Wee Tee Soh, National University Health System, Singapore
William Bao Bean, SOSV, China
Xavier Tadeo, N.1 Institute for Health, Singapore
Zia Zaman, Microsoft, Singapore
Notes
1. Rubin, “Epidemiology and Risk F actors for Spine Pain.”
2. Kaplan at al., Priority Medicines for Europe and the World: 2013 Update.
3. Wu et al., “Global Low Back Pain Prevalence and Years Lived with Disability
from 1990 to 2017.”
Introduction 17

4. Hartvigsen et al., “What Low Back Pain Is and Why We Need to Pay
Attention.”
5. Hajat and Stein, “The Global Burden of Multiple Chronic Conditions”;
Buttorff, Ruder, and Bauman, Multiple Chronic Conditions in the United States.
6. World Health Organization, “Cardiovascular Diseases (CVDs).”
7. Sherwin et al., “The Prevention or Delay of Type 2 Diabetes.”
8. World Health Organization, “Diabetes: Key Facts.”
9. Ahmad, Ormiston-Smith, and Sasieni, “Trends in the Lifetime Risk of
Developing Cancer in G reat Britain.”
10. World Health Organization, “Dementia: Key Facts.”
11. Centers for Diseases Control and Prevention, “Health and Economic Costs
of Chronic Diseases”; Lorenzoni et al., Health Spending Projections to 2030.
12. Hehner, Biesdorf, and Möller, “Digitizing Healthcare—Opportunities for
Germany.”
13. Sinsky et al., “Allocation of Physician Time in Ambulatory Practice.”
14. Haimovitch and Kirkner, “A Look B ehind Revision Optics’ Shuttered Door”;
Keown, “Sienna Biopharmaceuticals Files for Bankruptcy, Delaying Phase III
Psoriasis Trial”; Reuter, “Proteus Files for Bankruptcy: Where Did It Falter?”

• PART I
WHAT ARE DIGITAL THERAPEUTICS?
19

• Chapter 1
Digital Health Solutions versus

Digital Therapeutics
This chapter was written with contributions by Dr. Jessica Shull,

Head of Digital Therapeutics at Vicore Pharma in Sweden. Previously,
she led the European Policy work for Digital Therapeutics Alliance, the
global digital therapeutics (DTx) industry association.
Digital health and digital therapeutics are popular terms that are often
used interchangeably, yet critical distinctions between the two exist.
“Digital health,” as an umbrella term, encompasses all technologies,
platforms, and systems across the wellness and health care industries
that engage consumers for lifestyle, wellness, and health-related pur-
poses.1 It includes all trackers and “quantified self” apps, platforms, or
solutions, from such consumer wearables as sleep trackers and fitness
apps (e.g., Apple watch, Fitbit’s range of fitness products, and the Oura
Ring for sleep and activity tracking) to clinically evaluated cardiac
health monitors and cloud infrastructures (e.g., BioTelemetry, Coala
Heart Monitor, and Philips VitalHealth). U nder this concept of digital
health, we identify three categories of digital health solutions: well-
ness and support, diagnostic and monitoring, and digital therapeutics
(table 1.1).
21

Table 1.1. Three categories of digital health solutions
Wellness and support Diagnostic and monitoring Digital therapeutics
Overview Products that capture, store, Products that measure Products that deliver
and transmit health data and/or intervene therapeutic interventions
directly to patients
Clinical evidence Not typically required Required Required
Real-world Not typically required Not typically required Required
outcomes
Regulatory Varies depending on the Required Required
oversight intended use and function of
the solution
Examples Lifestyle apps and fitness Digital diagnostics, digital Digital cognitive therapy,
trackers, telehealth platforms, biomarkers, remote patient digital behavioral therapy,
health information technology, monitoring, medication digital physical therapy
consumer health information, adherence tools, ingestible
enterprise support sensors,1 connected drug
delivery devices
Source: Adapted from Digital Therapeutics Alliance, Ensuring Appropriate Quality, Access, and
Utilization of Digital Therapeutics .
1
What about wearables and sensors? Additional functionalities, digitally delivered or not, may be
required to optimize further patient care and health outcomes, such as wearables and sensors,
ingestible or not. T hese functionalities could be import ant to the physicians to assess and optimize
overall therapy (e.g., clinical decision support system or face-t o-face education and training). On their
own, t hese components c an’t be considered DTx, but they could be import ant elem ents to the DTx’s
overall value proposition and market success.
As an independent category within digital health, the appeal of dig-

ital therapeutics, or DTx, lies in blending patient-centric technologies
with evidence-based medicine to treat, manage, and prevent a broad
spectrum of chronic conditions and disorders. According to Click Ther-
apeutics, which develops and commercializes digital medical treat-
ments, the term “digital therapeutics” was coined in 2012, the same
year their company was founded (so they would know). Back then, it
was defined as “outcomes-based mobile software,” quite similar to its
current definition by Pear Therapeutics—another DTx player that de-
velops prescription-based digital interventions—as “software-based
disease treatments.”
The nonprofit trade association Digital Therapeutics Alliance (DTA)
characterizes DTx as follows:2
22 What Are Digital Therapeutics?

DTx deliver therapeutic interventions3 directly to patients . . . [that]
may be used independently or in tandem with remote or in-person
clinician-delivered therapy to optimize patient outcomes. . . . Digital
therapeutics undergo clinical t rials, collect real-world outcomes, and are
based on patient-centered core principles and product development best
practices, including product design, usability, data security, and privacy
standards. . . .
What value could DTx products bring to this situation?
• Increase remote access to therapies that are clinically demonstrated
as safe and effective.
• Provide care independent of a patient’s schedule and in the safety of
their home environment.
• Are easily scalable and often accessible through patient-owned
devices (e.g., smartphones, tablets).
• Generate actionable, real-world insights that enable intelligent
data-driven care management and clinical decision making.
Similarly, DTx must also adhere to ten foundational principles:4
• Prevent, manage, or treat a medical disorder or disease

• Produce a medical intervention that is driven by software
• Incorporate design, manufacture, and quality best practices
• Engage end users in product development and usability processes
• Incorporate patient privacy and security protections
• Apply product deployment, management, and maintenance best
practices
• Publish trial results inclusive of clinically meaningful outcomes
in peer-reviewed journals
• Be reviewed and cleared or approved by regulatory bodies as
required to support product claims of risk, efficacy, and
intended use
• Make claims appropriate to clinical evaluation and regulatory
status
• Collect, analyze, and apply real-world evidence and/or product
performance data
Digital Health Solutions versus Digital Therapeutics 23

In a nutshell, DTx are a subset of digital health that is based on clini-
cal evidence, generate real-world outcomes, and are subject to regula-
tory oversight.
How Do DTx Work in Practice?

To understand how DTx work in practice, let’s look at diabetes, one of
the top 10 causes of death globally,5 as an example. You probably don’t
need to be convinced about the burden diabetes represents to patients,
families, and our health care systems. Patients with type 1 or type 2 dia-
betes often struggle with managing their condition, and this is especially
true in the months a fter they are diagnosed, when the vast majority of
them experience difficulties with keeping their blood glucose level be-
low recommended thresholds.6 This is a problem waiting for a DTx
solution.
Granted, there are myriad of mobile applications with various fea-
tures for diabetes management that are now available, each offering a
mix of such functionalities as food diaries, calorie counters, and weight
management programs. A few of t hese apps, however, have been de-
veloped using evidence-based frameworks (appendix B) and would
therefore not be considered digital therapeutics. BlueStar®,7 a software
program powered by Welldoc and cleared by the US Food and Drug Ad-
ministration (FDA) as a Class II medical device for diabetes, is one of
the exceptions.
DTx Case Study: BlueStar®

BlueStar®, an AI-driven DTx solution available as a mobile app (fig-
ure 1.1) and a web version, offers users advanced coaching and tailored
guidance for managing diabetes, including medication administration,
physical activity, smart food choices, and psychosocial well-being.
Since its launch in 2013, the platform offers a robust set of capabil-
ities for type 1 and type 2 diabetes management. For example, a patient
can use the insulin dose calculator to find out the exact dose of bolus

Figure 1.1. Overview of BlueStar’s mobile app. Source: Photo by Welldoc Inc., 2022
insulin they should take for a given carbohydrates intake amount or

blood glucose value. The platform can also analyze and report blood
glucose test results and support medication adherence by providing
coaching messages (motivational, behavioral, and educational) based
on real-time blood glucose values and trends.
As a result, based on individual readings, patients are gently “nudged”
by a selection of more than 30,000 automated, tailored coaching
messages to take the most appropriate medication, diet, or exercise-
related action, driven by clinical and evidence-based guidelines.8 The
compatibility of BlueStar® with a number of blood glucose monitoring
and continuous glucose monitoring devices, blood pressure cuffs, and
weight scales further facilitates its seamless integration into a patient’s
day-to-day lifestyle.
Since using the BlueStar® app does not require a prescription, the
company is working closely with payers and employers so current and
future users can have access to the app directly through their health

plans and wellness programs—although a journey with BlueStar® can
also start with a physician recommending it.
In 2020, the FDA cleared BlueStar’s Insulin Adjustment Program
(IAP), a novel medical prescription (Rx) feature that supports individuals
with type 2 diabetes using long-acting insulin. Patients who are specifi-
cally prescribed IAP, in addition to the basic BlueStar® platform, are
guided through the process of basal insulin titration and benefit from
the ease of real-time adjustments to their insulin without having to use a
calculator. Insulin initiation and titration require a lot of effort and ad-
justments from patients and their health care providers, so this advanced
insulin calibration feature fills an important need. Concretely, health
care providers use a web interface to prescribe an initial dose and “Blue-
Star® does the rest.”9 It also allows physicians, diabetes educators, and
care coordinators to follow the evolution and progress of their patients
through their own dashboards and intervene in a highly targeted way
when a situation of risk has been flagged by the system. This is an exam-
ple of a digital therapeutic intervention delivered directly to patients.
Diabetes solutions with BlueStar® Rx is just one example of how
DTx can help patients in managing their conditions and health care
professionals in remote monitoring their patients. Fortunately, the
therapeutic areas, applications, and opportunities for DTx go beyond
diabetes and these technologies can be used in a range of clinical indi-
cations, including for pain, attention deficit disorder, m ental health
disorders, and cancer assistive therapy at various stages of develop-
ment and regulatory clearance (table 1.2).
According to their mechanism of action, DTx can be further subdivided

into two broad categories:
• DTx used as standalone treatments (a.k.a. drug-replacement

therapies) and
• DTx used in conjunction with other therapies (a.k.a. adjunctive
digital therapies, companion digital therapies, or drug-plus DTx)
to augment their effectiveness, such as BlueStar’s AI-guided
insulin adjustment program.

Table 1.2. Selected examples of well-k nown DTx solutions
Companies and
therapeutic areas DTx solutions Selected outcomes
Akili Interactive: Utilizing adaptive sensory stimulus Indicated to improve attention

Attention deficit software for the treatment of ADHD function as measured by
hyperactivity delivered through a captivating and computer-based testing.1
disorder (ADHD) engaging video game experience.
Big Health: Sleep Sleepio is a digital sleep improve- 76% of long-term poor sleepers
improvement ment program based on digital achieve healthy sleep.2
program cognitive behavioral therapy (dCBT)
to help overcome poor sleep and
thereby improve mental health.
BlueStar®: Diabetes BlueStar® combines individual A 1.7-to 2.0-point mean A1c
(metabolism) patient treatment data with machine reduction for adults living with
learning algorithms to provide type 2 diabetes who used
tailored motivational, behavioral, and BlueStar® as well as a savings,
educational coaching messages to on average, from US$254 to
aid in diabetes self-management. US$271 per user per month.3
Click Therapeutics: DTx-as-a-service solution builder for 45.2% of the participants
Smoking cessation biopharma companies, leveraging (n = 188) had stopped smoking
the company’s Click Neurobehavioral at the end of the study.4
Intervention Platform. It has a
pipeline of other DTx for a variety of
indications. Clickotine, the most
advanced one, is a clinically
validated fully digital smoking
cessation program.
Kaia Health: Delivery of physical exercises, Users report an average pain
Musculoskeletal behavioral therapy, and education for level decrease of 43% over
disorders chronic back pain patients, to help 12 weeks.5
both the body and the brain cope
with musculoskeletal conditions. The
company also has a DTx for
pulmonary rehabilitation.
MedRhythms: Neurologic music therapy to address A single, fully automated training
Neurologic disease motor, speech, and cognitive visit resulted in increased usual
and injury dysfunction caused by neurologic and fast walking speeds.6
disease or injury.
Pear Therapeutics: Prescription DTx to treat patients Patients in recovery from SUD,
Substance use suffering from a range of serious whose primary substance of
disorder 7 diseases, including substance use abuse was not opioids, were
disorder (SUD), opioid use disorder more than two times as likely to
(OUD), and chronic insomnia. abstain from drug use when
reSET was added to outpatient
therapy versus when it was not.8
(continued)

Table 1.2. (continued)
Companies and
therapeutic areas DTx solutions Selected outcomes
Voluntis: Cancer Oleena’s algorithms serve cancer More effective communication

patients by providing personalized between patients and health
guidance on how best to mitigate care providers; personalized
symptoms, using on-demand supportive care plan for better
directions for initiation and dosing adherence and outcomes.
of supportive therapies based on
the patient’s individual care plan. It
also provides patient monitoring
and generates insights for the care
team.
Note: DTx companies listed in alphabetical order.

1
Akili Interactive, “Akili Announces FDA Clearance of EndeavorRx for C hildren with ADHD,
the First Prescription Treatment Delivered through a Video Game.”
2
Espie et al., “A Randomized, Placebo-C ontrolled Trial of Online Cognitive Behavioral
Therapy for Chronic Insomnia Disorder Delivered via an Automated Media-R ich Web
Application.”
3
Welldoc Inc., “WellDoc Validates Potential of Its Digital Therapeutic, BlueStar, to
Significantly Reduce Healthcare Costs.”
4
Iacoviello et al., “Clickotine, a Personalized Smartphone App for Smoking Cessation.”
5
Toelle et al., “App-B ased Multidisciplinary Back Pain Treatment versus Combined
Physiotherapy Plus Online Education.”
6
Hutchinson et al., “A Music-B ased Digital Therapeutic: Proof-o f-C oncept Automation of a
Progressive and Individualized Rhythm-B ased Walking Training Program a fter Stroke.”
7
Before this book's publication, Pear Therapeutics began a reorganization.
8
Pear Therapeutics, “Reset for Recovery: The Proof.” Accessed November, 11, 2022.
https://w ww.r esetforrecovery.c om/r eset#the-p roof.
So how exactly do digital therapeutics differ from digital health over-

all? Below we outline three essential advantages that distinguish DTx
from the broad mix of digital health inventions.
DTx Differential Advantage #1: Clinical Evidence

at the Personal Level
Within the digital therapeutic concept, therapeutic is an important
word. It refers to the intended use of the digital solution and highlights
the fact that there is clinical grade evidence backing up its safety and

efficacy, two critical aspects of any therapy that are demonstrated
traditionally in randomized clinical t rials (RCTs). Health care pro-
viders weighing the option of recommending or prescribing a DTx
to their patients should always read carefully the product label,
whose importance is “paramount, because it’s the sacred communi-
cation that the company can make to the prescribing clinicians, and
it defines if the product is safe, efficacious and may provide benefits
to the patients,” suggests Corey M. McCann, president and CEO of Pear
Therapeutics.10
Indeed, the level of clinical evidence is of utmost importance h ere
as it is what differentiates DTx from other digital health solutions.
The latter kind may indirectly lead to clinical benefits, but they are
not subjected to clinical trials as per regulation by specific coun-
tries. On the contrary, Pear Therapeutics’s reSET and reSET-O first
products, cleared by the FDA, have been tested and validated in six
RCTs involving over 1,500 patients diagnosed with substance use
disorder.11
According to the levels of evidence used to rank the quality of med-
ical and health care research, RCTs and systematic reviews of RCTs
remain the gold standard for evaluating the effectiveness of interven-
tions, including DTx. This hierarchy provides guidance about the
types of evidence that are more likely to provide trustworthy answers
to the clinical questions we are trying to answer. For instance, if we
look at the potential treatment benefits of DTx in answering the
question “Does this intervention help?,” findings from an RCT w ill be
of level 2 (with level 1 being the highest and level 5 being the lowest),
while those from a systematic review of RCTs or of N-of-1 trials w ill
be of level 1.12
Although both RCTs and N-of-1 trials are necessary, some research
requires trial designs that are difficult to implement in practice or
that make difficult data standardization afterward. Suffice it to say that
these challenges are surmountable and even play a role in driving clin-
ical trial design innovation.

Randomized Clinical Trials (RCTs) vs. N-of-1 Trials
RCTs provide estimates of how well a treatment works on average for a group of
patients in a trial. However, since individuals are all different from one another, an
individual may experience benefits or side effects that are different from those
experienced by other trial participants or by the average trial participant. In short,
RCTs do not predict how a particular individual w
ill respond. By contrast, N-of-1 or
“individual RCT” trials are single-patient trials of a treatment’s effectiveness and
safety. Toward a more personalized medicine enabled by DTx, N-of-1 trials provide a
more individualized assessment of therapeutic benefit at the personal level.
DTx Differential Advantage #2: Real-World Evidence Generation

Evidence of the benefits of DTx may come not only from clinical t rials,
but also from what we call real-world evidence (RWE). By design, DTx
have the capacity to collect and analyze, in real or near-real time, ex-
tremely valuable real-world data (RWD) about a patient’s current state
of health. Such data provide researchers and health technology assess-
ment specialists with powerful insights relating to the usage, bene-
fits, and risks of an intervention that conventional drugs, devices, and
treatments usually can’t give. As any clinician w ill tell you, RWD are
crucial for treatment decisions, since they provide an additional un-
derstanding of treatment risks and benefits in routine clinical prac-
tice, in contrast to evidence generated u nder the highly controlled con-
ditions of clinical t rials.
To put this into perspective, we know that often clinical trial partic-
ipants are not representative of real-world patients.13 Sometimes this
may be due to cohort size (about 60% of t rials enroll 100 or fewer sub-
jects14), a lack of gender and racial diversity, overly restrictive eligibility
criteria (i.e., frequent exclusion of elderly patients and patients with
comorbidities), or selection bias that favors positive results (i.e., re-
cruitment of predominantly young healthy people). The implication of
such enrollment practices is that trials may exclude many patients seen

in everyday clinical practice, thereby limiting the applicability of RCT
findings to usual care settings and reinforcing the importance of RWE.
Real-world evidence is of greatest importance since clinical trials are only

as good as they have been designed and may not reflect the true effective-
ness of a solution in the real world. You have for example overly restrictive
eligibility criteria, such as t hose of a recent trial assessing a DTx, which
excluded patients who c ouldn’t commit to spending 40 minutes a day on
their smartphone for 7 days a week. That’s not what DTx are designed
for or how they are intended to be used or fit in patients’ day-to-day
life.15 —Ken Cahill, CEO and cofounder of SilverCloud Health
Due to the dynamic nature of the DTx development process in re-

sponse to new data, DTx are also particularly well-suited for leverag-
ing RWD to continuously improve how interventions are designed and
delivered. The aim of such iterative fine-tuning—sometimes achieved
by h uman engineers and sometimes by reinforcement machine-
learning algorithms—is to optimize clinical outcomes. In this sense,
the capacity of DTx to generate RWE complementary to evidence gen-
erated in clinical t rials and to feed it back to improve the technology
is particularly important. This is b ecause all stakeholders, from pa-
tients to physicians to regulators, are intrinsically interested in un-
16
derstanding how DTx can deliver value above and beyond conventional
therapies and other digital health applications.
DTx Differential Advantage #3: Regulatory Oversight

T here is a clear line separating how pharmaceutical drugs and supple-
ments are marketed and used. As a consumer, you surely would have
seen on supplements’ labels the message “This statement has not been
evaluated by the FDA. This product is not intended to diagnose, treat,
cure, or prevent any disease.” The reasons are s imple: supplements, un-
like drugs, don’t have to demonstrate their safety or effectiveness be-
fore reaching the market, don’t require post-marketing surveillance,
don’t follow pharmaceutical good manufacturing practices, and their

manufacturers are not allowed to make disease treatment claims. In
essence, as supplements are not intended for medical use, they are not
regulated in the same manner.17
Just as pharmaceutical drugs are distinct from supplements because
of stringent regulatory requirements they must meet before being
marketed, DTx are distinct from digital health solutions in general.
Namely, DTx are commonly regulated u nder the software as a medical
device (SaMD) framework developed by the International Medical De-
vice Regulators Forum. This means that DTx must comply with regula-
tory requirements for software medical devices throughout their entire
life cycle, starting from product development all the way to post-
market monitoring and surveillance following product introduction.18
Even still, the traditional paradigm of medical device regulation was
developed specifically with medical devices in mind and as such does
not fully cover the unique characteristics of DTx. This necessarily pre
sents limitations when using SaMD to evaluate such novel technolo-
gies. To respond to this new set of challenges and to the growing in-
terest in using DTx in clinical practice, national authorities and payers
have started introducing innovative regulatory approaches that w ill
likely eventually converge around new standards across countries. As
of 2022 we have observed several different approaches:
• In Singapore: Currently DTx fall u nder the definition of

medical devices and as such are regulated by the Health Sci-
ences Authority (HSA) u nder the Health Products Act.19 Con-
scious of the complexities and challenges of incorporating
emerging new technologies, such as artificial intelligence
applications intended to be used for medical purposes, the HSA
recently issued guidelines to provide clarity on the regulatory
requirements for software medical devices throughout their
entire life cycle from product development all the way to
post-market surveillance.20
Recently, Biofourmis, which is focused on predictive care,
obtained HSA approval for its two AI-based platforms: Biovitals

Analytics Engine, which enables constant monitoring of
individual patients’ physiology-based health data, and Rhythm
Analytics, which provides automated interpretation of cardiac
arrhythmias. The digital therapeutic reSET from Pear Thera-
peutics was also the first HSA-authorized prescription DTx to
be available in Singapore in 2020.
• In Germany: At the end of 2019, Germany launched its Digital
Healthcare Act, introducing the concept of “app on prescrip-
tion” as part of health care provided to patients. U nder this
new legislation, physicians and psychotherapists can prescribe
digital health applications (digitale Gesundheitsanwendungen, or
DiGA for short).21 To get on the list of approved and reimburs-
able apps, t hese solutions need to successfully pass the assess-
ment of Germany’s Federal Institute for Drugs and Medical
Devices. The entire process from filing to approval has been
designed to be a fast-track process that can be completed
within just three months. As of March 2023, over 500 DiGA
applications had been submitted to the institute for review, 28
DiGAs had been provisionally approved, and 16 had been
permanently approved.22 Following Germany’s lead, other
reimbursement pathways are emerging in Europe (e.g., France,
Belgium, UK, Italy).
• In the European Union: The eHealth Stakeholder Group, the
European Commission’s expert group on European digital
health policy, published Proposed Guiding Principles for Reim-
bursement of Digital Health Products and Solutions. This docu-
ment intends to help member states’ authorities “to modernise
national reimbursement and financing schemes for the purpose
of enabling digital transformation in EU healthcare systems.”23
• In the United Kingdom: In early 2020, the National Health
Serv ice (NHS) published a draft version of its Digital Health
Technology Standard (DHTS),24 which aims to accelerate how
digital health technologies are reviewed, commissioned, and
scaled for use across the wider health and care system, as well

as to provide guidance to digital health technology developers.
The DHTS focuses particularly on requirements and safeguards
that cover data-driven technology solutions, the development
and design of such solutions, and the regulatory framework that
ensures data protection and security. Public insurance coverage
is decided by local NHS organizations, such as the 43 existing
Integrated Care Systems. Additionally, the NHS has negotiated
licensing agreements with selected DTx developers, allowing
for free-of-charge DTx available in the NHS app library.
• In the United States: In 2017, the FDA launched the Digital
Health Software Precertification (Pre-Cert) Program as the
agency’s response to an increasingly complex digital health
ecosystem powered by fast-paced technological innovation. The
program aims to provide more streamlined and efficient
regulatory oversight of software-based medical devices devel-
oped by DTx manufacturers who have demonstrated a robust
culture of quality and organizational excellence, as well as
commitment to monitoring the real-world performance of their
products once they are commercialized in the US market. DTx
inclusion in corporative benefit packages and private health
insurance schemes is growing, and some Medicare and Medic-
aid programs guarantee public insurance coverage of DTx.25
hese regulatory advances reflect the multifaceted transformation of

T
traditional models of medical care, self-management, and prevention
propelled by DTx. For the time being, however, DTx organizations need
to navigate an uncertain and evolving regulatory space and manage
unclear business risks, such as the time required to bring a DTx to the
market, and challenges, such as those linked to reimbursement.
Earlier in this chapter while introducing digital therapeutics, we
emphasized the importance of the word therapeutic. Before we conclude,
we would also like to highlight the importance of the word digital.
DTx deliver evidence-based therapeutic interventions that aim to pos-
itively influence patients’ health status or behavior, with the primary

Key Takeaways
In addition to drugs and medical devices, physicians can now prescribe digital
therapeutic (DTx) solutions, either as a standalone treatment or in tandem with
current therapy options, to achieve improved health outcomes. As a category, DTx
are a subset of digital health, grounded in clinical evidence and compliant with
regulatory requirements. Due to their capability to deliver software-based disease
treatments (via a program, application, or platform), they should not be confused
with digital wellness and lifestyle apps, which cannot and do not make treatment
claims. A useful analogy that illustrates the difference between DTx and digital health
is the distinction between pharmaceutical drugs (FDA-cleared and -regulated) and
supplements (unregulated).
mechanism of action being performed digitally. In essence, DTx inno-

vations are made possible thanks to advances in technology with the
potential to change the way researchers assess clinical effectiveness
(chapter 7), the way regulators exercise regulatory oversight and, put
simply, the way health professionals provide health care.
In conclusion, DTx are by nature patient-focused solutions, de-
signed to be used directly by patients or their caregivers, such as par-
ents of children diagnosed with ADHD using Akili Interactive’s solu-
tion EndeavorRx. They offer the possibility to change the course of
disease throughout the care continuum, from diagnosis26 and early
detection to management and treatment of medical disorders, includ-
ing the possibility to optimize the effects of medication. Ultimately,
DTx aim to improve patient health outcomes by enabling more indi-
vidualized and continuous care, while leaning on the best available ev-
idence. In this book we w
ill walk you through examples and evolutions
that show how the combination at scale of technology with evidence-
based medicine has the revolutionary potential to transform person-
alized patient care.27

Notes
1. Digital Therapeutics Alliance, Digital Health Industry Categorization.
2. Digital Therapeutics Alliance, “Finding Value in Digital Therapeutics during
COVID-19.”
3. Intervention is the main reason why, in the potential absence of real-world
outcomes, diagnostics and monitoring differ from DTx. Solutions for symptoms
monitoring (e.g., COVID-19 symptom monitoring), medication adherence tools
(e.g., smart pill bottles or dispensers), or ingestible sensors (e.g., the now-defunct
Proteus) are not considered DTx as t hese solutions track and monitor but do not
aim to alter patients’ states or behaviors. T here is no intervention.
4. Digital Therapeutics Alliance, Digital Therapeutics Definition and Core
Principles.
5. World Health Organization, “The Top 10 Causes of Death.”
6. Fiagbe et al., “Prevalence of Controlled and Uncontrolled Diabetes Mellitus.”
7. BlueStar® and BlueStar® Rx are registered trademarks of Welldoc Inc.
8. Digital Therapeutics Alliance, “DTx Product Profile: BlueStar.”
9. Welldoc Inc., “Welldoc Receives FDA Clearance for Long-Acting Insulin
Support for Award-Winning Digital Health Solution BlueStar.”
10. Corey M. McCann, “Emerging Prescription Digital Therapeutics” (presenta
tion, APACMed Virtual Forum, Singapore, September 24, 2020).
11. Pear Therapeutics, “Pear Therapeutics Closes $20 Million Financing.”
12. OCEBM Levels of Evidence Working Group, “The Oxford 2011 Levels of
Evidence.”
13. Kennedy-Martin et al., “A Literature Review on the Representativeness of
Randomized Controlled Trial Samples and Implications for the External Validity of
Trial Results.”
14. Califf et al., “Characteristics of Clinical T rials Registered in ClinicalTrials
.gov, 2007–2010.”
15. Ken Cahill, “Emerging Prescription Digital Therapeutics” (presentation,
APACMed Virtual Forum, Singapore, September 24, 2020).
16. US Food and Drug Administration, Use of Real-World Evidence to Support
Regulatory Decision-Making for Medical Devices.
17. Although supplements do not need FDA safety and performance approvals
before commercialization, manufacturers and distributors are “responsible for
evaluating the safety and labeling of their products before marketing to ensure
that they meet all the requirements of the Federal Food, Drug, and Cosmetic Act as
amended by DSHEA and FDA regulations,” and the “FDA has the authority to take
action against any adulterated or misbranded dietary supplement product a fter it
reaches the market” (US Food and Drug Administration, “Dietary Supplements,”
accessed April 4, 2022, https://w ww.fda.gov/food/d ietary-supplements).
18. Health Sciences Authority, Regulatory Guidelines for Software Medical Devices:
A Lifecycle Approach.
19. Singapore Statutes Online, “Health Products Act 2007.”

20. Health Sciences Authority, Regulatory Guidelines for Software Medical Devices.
21. A Digital Health Application is defined by the Digital Supply Act (DVG) in
Germany as a medical device with the following characteristics: Medical device of
risk class I or IIa (according to MDR or, within the framework of the transitional
regulations, according to MDD); The main function of DiGA is based on digital
technologies; the medical purpose is essentially achieved by the main digital
function; The DiGA supports the detection, monitoring, treatment or alleviation
of diseases or the detection, treatment, alleviation or compensation of injuries or
disabilities; The DiGA is used by the patient or by the serv ice provider and the
patient together.
22. Roehl, “DiGA: Digital Therapeutics in Germany”; Bundesinstitut für
Arzneimittel und Medizinprodukte [Federal Institute for Drugs and Medical
Devices], DiGA-Verzeichnis [DiGA directory], accessed April 4, 2022. https://d iga
.bfarm.de/de/verzeichnis.
23. eHealth Stakeholder Group, Proposed Guiding Principles for Reimbursement of
Digital Health Products and Solutions, 5.
24. Joyce, NHS Digital Health Technology Standard.
25. Ravot and Ascione, “Access/Reimbursement Policies for Digital Therapeu-
tics Already in Use in National Health Systems”; Digital Therapeutics Alliance.
“DTx by Country.”
26. It’s import ant to note that according to the Digital Healthcare Act in
Germany, solutions aimed at primary prevention of disease are not considered DTx.
To be considered DTx, an application should contribute to preventing the worsening
of an already diagnosed disease (secondary prevention) or of a secondary disease or
complication (tertiary prevention).
27. Evidence-based medicine is defined as “the conscientious, explicit and
judicious use of current best evidence in making decisions about the care of individ-
ual patients” (Szajewska, “Evidence-Based Medicine and Clinical Research,” 14).

• PART II
THE RIGHT TOOL FOR THE

RIGHT PROBL EM
All too often in the technology world we encounter examples of a

technology solution looking for a problem. But the best and most
useful innovations—in health care or elsewhere—result from identi-
fying and defining the problem first and growing a solution from
there. If one is looking at a problem from a single vantage point (e.g.,
from a technologist’s or clinician’s perspective), it is easy to overlook
important aspects of building a thoughtfully designed and scalable
business while keeping the end user in sight at all times. As you, our
reader, may be e ager to get a promising DTx technology built, a first
pilot study r unning, or a first round of funding secured, there is a
risk you may be so focused on operational details that you are missing
the forest for the trees. In this first part of the book, we delve into
the details of venture formation. We offer you three perspectives to
help you make sure you are solving the right problem and equip you
with the knowledge to overcome common challenges faced by DTx
organizations at the initial stage.
Checklists save lives. That’s the powerful message the World Health
Organization championed back in 2008 to encourage surgeons across
the world to use checklists when they operate as a way to guarantee
safety and avoid costly m istakes.1 Checklists can also save emerging
organizations from pitfalls that can cost them their businesses. As
you gear up for your DTx development journey (or even if the idea is
still just a seed in your mind), we invite you to consider the items on
the following checklist. They are intended to serve as a balcony from
which you can view and assess the progress of your venture down on
the ground, preemptively recognize challenges that lie ahead, and take
39

timely or corrective action as you advance your idea. Building a DTx is
an iterative process, so be sure to revisit this checklist as needed, as
well as o
thers you w
ill find in the next chapters.
❒ What is your DTx organization’s primary purpose and objec-

tive (a.k.a. its “true north”)?
❒ Who are the most influential stakeholders in your ecosystem
(i.e., patients, physicians, partners, or payers)?
❒ What are the jobs each of t hese stakeholder groups is trying
to get done?
❒ Who are your target users, purchasers, and influencers?
❒ What is your systematic, repeatable approach to analyzing
stakeholder needs?
❒ How do you deliver exceptional utility along the care
continuum?
❒ How w ill your organization deliver annual revenues by a
targeted time frame?
You should be able to answer all of t hese questions once you have a
clear view of the problem your DTx technology aims to address. A
problem well defined is a problem half-solved, right? Using your
problem definition as your true north during every stage and pivoting
along your organization’s strategy is the most direct way to success.
All too often, though, developers and entrepreneurs operate in
reverse: they have a solution looking for a problem, obviating whether
the solution is actually needed. Indeed, the absence of a clearly defined
market need has always been one of the main drivers of startup failure
in any industry, not just in health care.2
Our interviews with entrepreneurial health care academics and
physicians have yielded key barriers to innovation success. For exam-
ple, it was noted that health care innovators tend to underestimate
early on how a pipeline program and its objectives may adversely
impact the viability and scalability of their technology and organ
ization l ater on. They recognized that major curveballs can often
come from erroneous assumptions or insufficiently thought-t hrough
40 The Right Tool for the Right Problem

ideas about the business model (e.g., direct-to-consumer approach
or not), the regulatory perspective (e.g., requirements for regulatory
clearance), or the reimbursement pathway. T hese outcomes can be
particularly unfortunate when the technology would in fact be
transformative had it been deployed optimally.
The initial steps of forming a DTx organization are indeed critical.
Many health technology success stories start with insights from health
care professionals, patients, and caregivers to understand their
challenges and what potential solutions may look like from their
points of view. Even if physicians or patients don’t know what they
want, cannot visualize what they want, or want a solution for a very
specific unmet clinical need that is not addressable at scale—that
information contributes to the broader context of your idea’s fit.
Where applicable and accessible, perspectives from providers, insur-
ers, and other key stakeholders may be vital resources for successful
emergence out of the ideation stage.
So the absence of a well-defined market need is a top reason why
startups fail, but r unning out of cash or not having the right team
in place are also critical risks. To address t hese challenges, DTx
developers need to identify the right problem across three dimensions
(see figure). We will explore each of t hese dimensions in the following
Ensuring business
viability
? Problem
Statement
Addressing unmet Addressing unmet

stakeholder needs clinical needs
Figure Part II.1. Identification of the (right) problem. Source: From the authors
The Right Tool for the Right Problem 41

chapters and offer perspectives, tools, and best practices that w ill
help make your venture formation and technology development road
map a success. T
hese suggestions further aim to help innovators
systematically approach roadblocks, ensuring they are asking the
right questions at the outset, and ultimately derisking e very step of
DTx development.
Notes
1. Humphreys, “Checklists Save Lives.”
2. CB Insights, “The Top 12 Reasons Startups Fail.”

• Chapter 2
Addressing Unmet Stakeholder Needs
This chapter was written with contributions by Fabien Souillard, a

patient in France living with type 1 diabetes; Dr. Ester Yeoh, senior
consultant at the Diabetes Centre, Admiralty Medical Centre and
Division of Endocrinology at Khoo Teck Puat Hospital in Singapore;
and Dr. Raj Kumar, certified Blue Ocean Strategy practitioner and
CEO of UCSI Blue Ocean Strategy Consulting in Malaysia. Their
perspectives are reflected in this chapter to guide you through the
exercise of identifying the right problems to solve with DTx.
Fabien was diagnosed with type 1 diabetes when he was 13 years old.
The concern that something was wrong started when he began to feel
frequently tired and was constantly thirsty. As a result, he began to
refuse to go outside his house to play with friends. His f amily became
increasingly worried and were prompted to bring him to the hospital,
where the diagnosis was made. At first Fabien didn’t comprehend fully
what it all meant and how it would impact his lifestyle, but from that
point on his life changed. He was started on daily insulin injections
and studied how to monitor his blood glucose and carbohydrate intake.
Ten years later, after countless consultations with endocrinologists
43

and other physicians, and after countless visits to the hospital to man-
age his double-digit hemoglobin A1c (HbA1c), stomach cramps, and
other complications arising from his condition, he still struggles with
daily finger pricks, carbohydrate counting, and trying his best to have
the semblance of a “normal” life. Like most p eople with diabetes, Fa-
bien has not asked for much. He has simply wanted to feel better while
minimizing the effort he must put into managing his condition.
Until recently, Fabien had never heard about Omada Health, Blue-
Star®, or any other diabetes-related DTx platform or app. Moreover,
the first time diabetes patients like him were on the verge of experi-
encing a revolution in their diabetes management was in 2006 when
the first inhaled insulin—Exubera—was approved by the FDA. Exu-
bera was a dry powder form of rapid-acting insulin delivered by oral
inhalation and labeled (which in regulatory language specifies an ap-
proval for a specific condition) for controlling hyperglycemia in adults
with type 1 or type 2 diabetes. According to Pfizer, which developed
the formulation, Exubera was a major medical and technological break-
through that promised to replace bothersome insulin injections.
It turned out, however, that not many patients and physicians were
interested in using Exubera. After more than a year on the market, de-
spite heavy advertising, Exubera only managed to capture 1% of the
insulin market. For this reason, Pfizer ended its production in late
2007 and took a $2.8 billion pretax hit on the product. “Despite our
best efforts, Exubera has failed to gain the acceptance of patients and
physicians,” acknowledged Pfizer’s CEO at the time.1 All things con-
sidered, Exubera highlighted the importance of understanding how pa-
tients, physicians, and payers approached diabetes management.
From a patient’s perspective, the product was challenging to use dis-
creetly, as it was administered through an inhaler about the size of a
can of tennis balls. Patients who w ere willing to use Exubera daily w
ere
required to check their lung function at regular intervals as a safety
measure. Also, it did not provide an easy way to convert a patient’s nor-
mal insulin amount into an Exubera-calibrated dosage, leading to

concerns about potential dosing errors. As a result, it created consid-
erable operational hurdles and complexities for patients, but also for
physicians and the system.
From a physician’s perspective, implementation was challenging.
Substantial effort was required to educate patients on how to use the
Exubera inhaler adequately—a deal-breaker for many physicians con-
sidering the limited consultation time available in a busy practice. For
instance, in the United Kingdom a general practitioner spends an aver-
age of 9.2 minutes with a patient, which is clearly insufficient for any
pedagogical activities in addition to the clinical consultation.2 Also, Ex-
ubera did not sufficiently address the issue of conventional insulin ad-
ministration, as most type 1 patients and a portion of type 2 patients
still needed long-acting insulins for basal control of blood sugars.
From a payer’s perspective, looking particularly at the cost-
effectiveness, the device was 30% more expensive than traditional
solutions. Bloomberg at the time concluded that “perhaps next time
Pfizer w ill do a l ittle more homework to determine w hether an inno-
vative idea truly has a market.” 3
The example of Exubera, a well-known case study in the industry,

reminds us that (1) the path from research and development to com-
mercialization is anything but straightforward, and (2) obtaining an
FDA approval does not guarantee market adoption. Yet too many organ
izations and entrepreneurs focus solely on their technology instead of
the problem it aims to solve. This condition is referred to sometimes as
entrepreneurial myopia, which is when companies are more focused on
their product than the needs and requirements of customers.
In the case of Exubera, understanding the problems faced by pa-
tients like Fabien, physicians, and other stakeholders involved di-
rectly or indirectly in the care of people with diabetes may have pre-
vented entrepreneurial myopia, or corporate myopia. More broadly,
prevention should start with diagnosing a potential technological solu-
tion’s ecosystem—which is typically dynamic and complex in health
care—and its direct and indirect stakeholders (figure 2.1), all of whom
Addressing Unmet Stakeholder Needs 45

• Employers • Individual patients
• Unions • Caregivers
• Government-run agencies • Friends and family
• Insurers • Patient advocacy groups
• Pharmacy benefit managers PAT
S
ER IEN
AY
TS
Digital
therapeutics
PA R T N E
ICIA N S
• Charities, social enterprises, • Allied health professionals
ecosystem
and voluntary groups • Hospital administrators
• Education/research community • Nurses/care teams
YS
RS
• Investors • Primary care physicians
PH
• Large tech companies • Pharmacists
• Small and medium-sized enterprises • Providers
• Pharma/medtech companies P L AYE R S • Specialists
• Policymakers
• Potential competitors and
alternative solutions,
complementary or not (e.g., drugs,
DTx, digital health solutions, etc.)
Figure 2.1. The 5Ps in the DTx stakeholder ecosystem. Source: From the authors
have different roles and objectives. With the rapidly evolving ecosys-
tem at the intersection of health care and technology, DTx teams must
be much more externally focused than any other organization and not
overlook the needs and goals of any stakeholder group, as they could
later turn out to be key catalysts for entrepreneurial success as well as
impact to the patient and health care system.
T here are numerous ways to slice and dice the stakeholder ecosys-
tem to gain a deeper understanding of how the diff erent agents inter-
act and depend on each other. Key f actors for t hese stakeholders to
consider when assessing a pending ecosystem include, but are not
limited to,
• evaluating each stakeholder’s potential impact;
• understanding their awareness about and interest in adopting
DTx solutions;
• the stakeholders’ level of influence over o
thers who might
adopt the solution; and
• the ability of DTx innovators to influence these stakeholders.

To use a health care analogy, performing a diagnostic analysis of the
DTx stakeholder ecosystem should be as thorough as a patient’s first
consultation with a specialist. T hese first consultations are typically
longer and more extensive as specialists take the time to understand
the patient’s medical story, check their medical records, understand
what m atters most to them (oftentimes patients value better func-
tional health more than mere improvement of clinical indicators),
and determine how they can help. Similarly, we are suggesting that se-
rious DTx entrepreneurs should begin by conducting a thorough ex-
amination and diagnosis of the ecosystem they aim to impact. Doing
so goes beyond traditional stakeholder mapping and extends to un-
derstanding the challenges and priorities of patients, physicians, and
other end users along the care continuum relevant to their specific
technology, from disease prevention to diagnosis and management. In
the process, they w ill likely discover that success looks very different
to each of t hese stakeholders and their needs could sometimes be con-
flicting, which is why it is important to take stock of all t hese inputs
before designing or trying to implement anything.
DTx has the capacity to transform the way that medicine is prac-
ticed if solutions are developed and implemented the right way. While
a whole-of-community approach among stakeholders to collaborate
and address this new frontier is underway, a historical view of health
care innovation and the realization of established adoption strategies
offer valuable pointers to t hose seeking to disrupt the field of medi-
cine. For this reason, in this chapter we offer a deep dive into diabetes
as a practical exercise of ecosystem diagnosis, sharing the perspectives
of Fabien, who has lived with diabetes for more than half his life, and
Dr. Ester Yeoh, a physician who recognizes the challenges faced by the
myriad of patients from all walks of life that she manages in the clinic.
Fabien and Dr. Yeoh’s stories may be unique, but they do have facets
of similarity to other stakeholders in the health care ecosystem. Both of
them would benefit from a potential DTx solution, either coming from
the angle of the person with diabetes or the physician managing that
person. Ideally, a DTx solution would be able to satisfy both needs.

Understanding their perspectives on managing diabetes and look-
ing at their current approaches to technology adoption w ill provide
insights into how DTx solutions can fit within their respective routines.
In the journey to creating a DTx solution, too, t hese perspectives w
ill
need to be referenced e very now and then to ensure that the focus
on the end user is not lost. Let’s start with looking at the challenges
faced by p eople living with diabetes. With t hese challenges comes
the opportunity for the ideation of effective and adoptable DTx
solutions.
Challenges Faced by P
eople Living with Diabetes
Delay in Diagnosis
Diagnosing diabetes is relatively simple in theory. Depending on
country-specific guidelines, a single blood test—either an HbA1c test,
a fasting plasma glucose test, or an oral glucose tolerance test—could
confirm a diagnosis. Singapore is frequently cited as having one of
the most efficient health care systems worldwide.4 Nonetheless, 50%
of diabetes cases in Singapore remain undiagnosed, mirroring global
estimates of 45.8% undiagnosed, ranging from 24.1% to 75.1% across
geog raphical regions.5 Increasing awareness of screening locations,
improving access to and compliance with clinician follow-up appoint-
ments, and harnessing technology are just a few of the many ways to
address this issue.6 Diabetes, including its diagnosis, is a global prob
lem of pandemic proportions, posing challenges to even the most well-
funded and efficient health care systems.7 It is imperative to overcome
this barrier to diagnosis b ecause diabetes tends to be the starting point
for many other health complications, and if not properly diagnosed and
managed, it can lead to cardiovascular, vision, renal, neurological,
gastrointestinal, and oral health disorders, to name just a few. In a
nutshell, its impact on patient quality of life can be pervasively
devastating.

Lack of Awareness and Short-Term Results
Public health messages about diabetes have been widely featured on
media channels as well as subway and bus stations ever since Singa-
pore initiated its war on diabetes campaign. However, there remains
a segment of the population that lacks awareness of the symptoms of
diabetes as well as the burdens of diabetes on everyday life due to the
additional daily tasks required to manage this condition. Once this
reality sets in, subsequent challenges in keeping diabetes u nder con-
trol quickly becomes a top-of-mind concern for patients, overriding
most other priorities. Fabien says, “Injection, diet, exercise . . . it is like
brushing your teeth. It’s about eating healthy, exercise, medication, you
have no break. You need to be cautious all the time. It’s particularly
frustrating when you think you have done everything right and your
blood sugar is still high. It takes a lot to stay u
nder control.”
From a patient’s perspective, despite the effort required, the reward
for the work put into effectively managing diabetes could simply be the
absence of symptoms or the absence of complications (such as hypo-
glycemia, which is low blood sugar as a result of pharmacological ther-
apy for diabetes).8 Yet the “punishment” for managing it poorly can
appear minimal at first; serious complications tend to materialize over
time and are thus often underestimated in the present. This can be
particularly apparent for younger diabetes patients, who often have an
invincible mindset, limiting their ability to assess risks and future con-
sequences.9 Fabien and his f amily w ere frustrated with the constant
need to be vigilant without seeing payback for their efforts, so they
began to look for more tangible, short-term results that would moti-
vate him to adhere to his treatment routine. “It’s more about how I look
now than about worrying about a potential kidney problem I might
have in ten years,” he says. The daily struggle to keep his blood sugar
and other indicators in check was resulting in an emotional roller
coaster for him. In disease management speak, it was resulting in incon-
sistent motivation and poor engagement.

For Fabien, it was necessary to start taking insulin immediately
after the diagnosis of type 1 diabetes. For people with type 2 diabe-
tes, though, insulin intake usually starts when controlling the disease
cannot be achieved with diet, exercise, and oral medications alone.
When a person with type 2 diabetes reaches that stage, they may view
it as a sign of failure, which is a common misconception.10 Further-
more, a successful initiation of insulin therapy may be challenging
due to medication avoidance from the fear of self-injection and per-
ceived social stigma.
Financial Burden
Individuals with diabetes tend to cover a substantial amount of out-
of-pocket costs, which can range from US$143 to US$2,210 per year.11
T hese costs include payment for consultations, medications that may
include insulin and insulin delivery devices such as insulin pumps, and
blood glucose monitoring equipment such as testing strips or glucose
sensors. T hese costs can impede or even preclude access to treatment,
which can subsequently lead to suboptimal patient outcomes.
The financial burden of diabetes extends beyond direct health care
costs, too. Indirect costs such as those caused by increased absentee-
ism, reduced productivity at work, or inability to work as a result of a
disease-related disability add to the burden, both for the person with
diabetes and for society at large.
Challenges Faced by Physicians Caring for Patients

with Diabetes
In order to make fully informed decisions when it comes to therapy
options, potential referrals, and counseling, physicians need to ac-
knowledge the full spectrum of challenges that lead to suboptimal
health outcomes for their patients. Yet, they also have challenges of
their own when it comes to supporting p eople with diabetes.

Time
Time is a major constraint for Dr. Yeoh and for all the health care practi
tioners we interviewed for this book, be it specialists, general practition
ers, or nurses. All mentioned how they wished they could have more
time with their patients during consultations so as to fully understand
their situations, needs, and potential obstacles to treatment. They all
highlighted that a one-size-fits-all approach does not work for people
with diabetes because each patient is affected by it in different ways.
Importantly, their expectations from the treatment results also vary
substantially. Therapeutic options need to be individually tailored and
collaboratively arriving at a joint decision requires time.
Diabetes management at a physician’s office is a multifaceted pro
cess. The physician needs to look through blood glucose records and
modify pharmacotherapy while also managing all risk factors simulta
neously. This includes monitoring blood pressure, cholesterol levels,
cardiovascular risk, smoking history, m ental well-being, and more. The
physician also ensures and coordinates screening for diabetes compli-
cations of the eyes, feet, nerves, and blood vessels as well as any addi-
tional problems that may surface during each consultation. It may not
be possible for a single consultation to encompass everything that re-
quires discussion, much less follow up on particulars like blood glucose
monitoring records. This results in additional out-of-clinic consulta-
tions through emails, phone calls, and other modes of contact. These
complex processes have even been considered as factors disrupting
work-life balance, potentially increasing the risk of physician burnout.
Low Uptake of Diabetes Self-Management Education

Across the landscape of diabetes management, Dr. Yeoh is most
passionate about empowering her patients to take charge of their con-
ditions through self-management education. Most health care prac
titioners and clinics that care for diabetes patients, however, do not

receive additional reimbursement for time spent providing nonclini-
cal serv ices. This misalignment of incentives in health care is neither
new nor unique to the therapy area of diabetes and is sometimes re-
ferred to as “unfavorable economics.”
Adding to the time constraint on the provider’s end, having diabe-
tes patients come for regularly scheduled educational sessions requires
effort and perseverance. The sign-up rates for such programs gener-
ally tend to be low and dropouts can reach up to 50%, increasing with
each additional day in the program. The reason for t hese poor atten-
dance numbers is that most patients cannot easily take time off from
work or school to attend activities that do not appear to be strictly re-
lated to clinical management of their condition, in contrast to the
blood sugar, eye, and foot screenings for which they already commit
substantial time. Dr. Yeoh has tried to work around t hose limitations
by organizing weekend and online classes but concluded that it’s an
uphill battle to convince patients to show up consistently when they
may not prioritize it themselves.
Confirming Dr. Yeoh’s observations, a team of researchers in 2017
investigated factors influencing uptake of structured diabetes self-
management education (DSME) by people with type 1 diabetes in or-
der to understand why only about 27% of t hese patients participated
in a common program known as the Dose Adjustment for Normal Eat-
ing program.12 They found associations between attendance at struc-
tured education and young age, female gender, and higher HbA1c level
(i.e., >75 mmol/mol doubled the chance of attendance). Also, partici-
pants from areas of social deprivation13 were half as likely to attend.
In 2019, these researchers were part of a team that identified five
key barriers for attending DSME: psychological capability (the most
influential f actor), numerical capability (defined as a person’s ability
to deal with numerical information, particularly in relation to their
diabetes self-care), internal/external judgment, thirst for knowledge,
and self-confidence.14 These themes led to the identification of distinct
profiles (also referred as “personas”) of p eople with type 1 diabetes
with low DSME attendance. This stratification approach can be lever-

aged by physicians, designers of patient education programs, and DTx
developers to define targeted interventions for specific patient sub-
groups, as the researchers noted. Could there be a DTx solution that
enables diabetes education to be brought to the patient seamlessly
while incentivizing the patient toward diabetes self-management?
Therapy Nonadherence
Despite a physician’s best efforts, people with diabetes may not con-
sistently follow their advice. Therapy nonadherence encompasses
skipping prescribed medication, forgoing necessary testing, and not
making recommended lifestyle changes. It is estimated that approxi-
mately 50% of patients do not take medications as prescribed.15 The
Diabetes Attitudes, Wishes, and Needs Study in 2001 showed that ad-
herence among type 2 diabetes patients was at 78% for medications,
64% for self-monitoring of blood glucose, 37% for diet, and 35% for
exercise.16 T here is a high correlation between nonadherence and low
socioeconomic status, low education levels, being a member of an eth-
nic minority, and depression, pointing to issues of inequality and in-
equity in diabetes management. Importantly, these are also factors to
consider in the context of DTx development.
The reported average adherence to long-term chronic illness ther-
apy in developed countries is 50%, implying a similar percentage for
nonadherence, according to the World Health Organization.17 It’s re-
ported that about 40% to 50% of diabetes patients in the United States
abandon treatment during insulin initiation or intensification, a per-
centage that can go as high as 70% to 80% in self-pay markets such as
India. In all cases, abandoning diabetes treatment contributes to worse
health outcomes.18
We know that therapy adherence is a key determinant of the effec-
tiveness of a treatment, but it can also have a positive economic im-
pact.19 In one example, the estimated cost savings of the US health care
system associated with improving medication adherence for diabetes
patients ranged from $661 million to $1.16 billion.20

Therapy nonadherence can be explained in many ways: inconve
nience, a low sense of urgency if symptoms are not immediately notice-
able or worrisome, low or no awareness about why behavioral change
matters, low perception of the risks of nonadherence and the benefits
of adherence, and a lack of immediate feedback to patients who do
comply with their treatment regimen regarding measurable progress.
In a nutshell, for people with diabetes, there is no apparent timely pos-
itive return on investment for their personal expense and efforts.
Behavioral Change
The insights from patient and physician have raised a clear point:
Behavioral change is at the foundation of a strong arsenal against
diabetes.
Instead of viewing diabetes as a constant battle with wins and
losses, Fabien has learned to live with it. Ideally, he wants to feel well
without his condition overwhelming his daily routines or requiring
drastic changes to his lifestyle. Unfortunately, standard treatment re-
quires substantial effort on the part of the patient. This reality is un-
likely to satisfy his needs and his personal diabetes goals, resulting in
a mismatch of expectations between physician and patient. All things
considered, the decisive factor is still Fabien’s commitment to adher-
ing to his treatment plan. T hose living with type 1 and type 2 diabe-
tes share this struggle.
Numerous studies have defined and assessed different behavior
change techniques, but identifying which specific techniques or com-
binations thereof are potentially effective for a given individual and
selected behavior presents a major challenge because behavior itself
is not a constant.21 A patient on Monday morning may be physiologi-
cally diff erent, feeling and acting differently, than the same patient on
Friday evening. Dynamically managing patient care at scale, then, is
currently an enormous challenge, even though studies show that there
is a clear correlation between patient behavior change and clinical out-
comes, particularly in diabetes.22

Complex is probably the word that best encapsulates Fabien’s and
Dr. Yeoh’s experiences in managing diabetes and dealing with its as-
sociated challenges of awareness, access to treatment, cost, time, and
motivation. T hose f actors highlight that the difficulties of managing
the disease arise not only from its clinical dimensions, but also from
the h uman dynamics around them: the patients, families, caregivers,
and health care professionals involved in the direct care of diabetes,
as well as the individuals and institutions who have an indirect im-
pact, including public and private payers, employers, technology in-
novators, and clinician-entrepreneurs who recognize the need for
better therapy adherence design. In the context of DTx, the objective
is to improve Fabien’s quality of life with a seemingly effortless man-
agement plan. As for Dr. Yeoh, the objective is to improve glycemic
control and lowering glycemic variability to realize positive clinical
outcomes while ultimately meeting a patient’s expectations and defini-
tions of success.
Pointer
For a 360° view of the DTx ecosystem, including perspectives other than those of
patients and clinicians, consider also understanding the perspectives of new
groups of relevant stakeholders, such as health care administrators, policymakers
(e.g., government and local authorities), payers (e.g., health system organizations,
employers, and insurers), academic researchers, investors, or industry players (e.g.,
biopharmaceutical and medtech organizations).
Diff erent stakeholders have unique views of how successful disease

management is defined (table 2.1). Their needs and interests could
sometimes conflict. Understanding their priorities, challenges, and
how they might be impacted is therefore critical to developing poten-
tial solutions. For a systematic, repeatable approach to analyzing stake-
holder needs, we detail in the next section of this chapter a theory of
consumer action, describing the mechanisms that lead market partic-
ipants (in our case, health care stakeholders) to adopt an innovation.

T able 2.1. Illustration of key patient and clinician definitions of success and
challenges using a DTx lens
Defining success in the context of DTx Potential challenges
Patients (plus • Feeling better and having fewer • Being overwhelmed by an

caregivers symptoms as a result of DTx-mediated overabundance of apps and
or family intervention. solutions available on the market,
members of • Spending less time on managing one’s and knowing how to choose the
patients) condition while having access to right one.
continuous, real-time, digitally generated • Feeling digitally illiterate
feedback about current health status. (a frequent concern for older
• Navigating and operating the DTx with users)
ease (e.g., simple to access; easy to use • Inaccessible infrastructure for
in-app charts, graphs, and other pictorial large segments of the population.
displays). • Concern about data protection
• Experiencing less financial burden. and privacy.
Clinicians • Empowering and supporting patients to • Lacking robust evidence of the

adopt healthier lifestyles likely to lead to DTx’s effectiveness (e.g., What
improved outcomes (e.g., improving are the proven clinical benefits?
patient awareness about their condition Can DTx really change patient
and the importance of therapy behavior?).
adherence). • Adding overhead and ending up
• Access to digitally generated intermedi- spending too much time on
ate outcomes feedback to aid with nonclinical aspects of care.
patient management. • Siloed implementation of
• Efficient patient onboarding with a existing technological solutions,
treatment plan. which could impede clinical
• Augmenting care delivered outside the workflow.
clinic by adding digitally enabled • Concern about accountability
touchpoints without investing additional (Who is to blame if the DTx
effort or time. provides the wrong advice?)
• Engaging the DTx with ease (e.g., simple • Concern about data protection
to access; easy to use and understand and privacy.
in-app charts, graphs, and other pictorial • Lacking access to the appropriate
displays). reimbursement models.
Did You Know?

In the last eight minutes you were reading this book, the US government has spent
close to $5 million ($4,977,169 to be exact) on diabetes alone, and this is only for
patients who have been diagnosed.23 We will see in the next few chapters how DTx
can contribute to having more informed, engaged, and motivated patients for better
clinical and financial outcomes.

Defining the Jobs Stakeholders Want to Get Done
Correctly identifying customer or stakeholder priorities and staying
focused on them when developing a new product or serv ice is essen-
tial to the success of any business, including a DTx startup. Here we
offer a systematic approach to understanding stakeholder priorities
and unmet needs by using the jobs-to-be-done (JTBD) framework.
In essence, we are all trying to get a job done. When we buy a
product, we “hire” it to help us do a job, as described by the theory’s
author, the late Clayton M. Christensen, who is widely regarded as
one of the foremost experts on business growth and disruptive in-
novation.24 In this line of reasoning, you may use a drill (product)
not to make a hole in the wall, but to hang a frame (job). Similarly,
you may buy the latest Bluetooth headset (product) not to listen
to music, but to motivate yourself and set your pace when you go
r unning (job).
In our collective experience, we have often seen the difficulties pa-
tients, families, and physicians have in clearly articulating their needs.
This puts an onus on you, the DTx innovator and creative problem-
solver, to figure out what would deliver value (defined differently to
each stakeholder) to the collective “them” (which requires a multifac-
eted vision). Henry Ford, founder of the Ford Motor Company, is known
for having famously said, “If I had asked p eople what they wanted, they
would have said faster horses.” DTx developers can borrow a page from
Ford’s book and apply an outcome-focused lens to overcome the chal-
lenge of vaguely framed problem definitions. The JTBD framework cre-
ates a unique value proposition for each stakeholder in the ecosys-
tem, so they can “get the job done.” Simply put, it’s ensuring that your
future solution w ill address a real problem.
Adapted from the literature, a JTBD analysis is generally structured
as follows:
[Person] wants to [Solve a Problem] in this [Context], so he/she can

[Expected Outcomes25]

Let’s see now how this works by applying it to Fabien’s case, our young
protagonist with type 1 diabetes. Fabien was growing impatient with
the lack of instant feedback or positive reinforcement offered by con-
ventional disease management programs. A DTx entrepreneur might
start by applying the JTBD framework to understand why and how Fa-
bien does what he does and the problems he f aces that motivate him
to buy and use a product. The aim is to understand Fabien’s priorities
in a particular set of circumstances, the events or situations that trig-
ger a reaction in him, his desired prog ress, and his motivation for
achieving that desired state. For example:
Potential Job #1:

Fabien wants to know what to do (e.g., exercise, monitor his blood glu-
cose, or other indicators, etc.) and what to eat when he is at home alone
so he can be confident that he is robustly managing his condition.
This first JTBD definition can lead to a solution, digital or not, that
aims to minimize the burden of self-management by simplifying the
treatment regimen (e.g., by delivering brief educational “nuggets” on
how to prevent blood sugar spikes). For Fabien to feel more empow-
ered, such a solution may take on many forms. Examples include dig-
itally delivered continuous feedback that gives him visibility into how
he’s d
oing in terms of therapy adherence, and perhaps positive rein-
forcement when he’s hitting all his goals. Obviously, the solution need
not be delivered only during Fabien’s regular outpatient visits to the
hospital. They can also follow him home, allowing him to stay con-
nected, see tangible results (and reasons) for him to be compliant, and
maybe even share his progress with friends or online peers to enable
community-driven adherence modalities.
Now notice how this next example can lead to a diff erent, more clin-
ically focused solution:
Potential Job #2:

Fabien wants to feel less bogged down by managing his type 1 diabetes
so he can have more time to do fun activities with his friends and family.

This second JTBD definition has to do with therapy optimization
and insulin titration, possibly facilitated remotely by a health care pro-
vider. It could materialize as a DTx solution that sends an alert to
Fabien’s treating physician to inform them if it seems like a consulta-
tion with a psychologist might benefit the patient at a particular point
in time, or if another kind of coaching or motivational program might
help with symptom management.
From our experience with designing new products that patients
want, there are five sets of questions that should guide this JTBD dis-
covery process:
1. What kind of problem is the patient trying to solve and in what

specific circumstance are they trying to solve it?
2. Is there an important problem they have that lacks an adequate
solution?
3. How is the patient getting the job done now? What are their
compensating behaviors in the absence of an optimal solution?
4. What functional, social, and emotional criteria w ill the patient
use to evaluate w
hether a potential solution is right for them?
Conversely, what functional, social, or emotional factors would
take a potential solution “off the t able” for them?
5. How would “value” be measured from the patient’s perspective?
We asked Fabien to share his perspective on each of these questions.

Please remember that Fabien’s story is as unique as any other patient’s
story, and we are not pretending that by understanding him better you
w ill be able to design the right DTx solution for all p
eople with type 1
diabetes. The overarching objective is to instead understand the
process and tools that you w ill need along this discovery journey. Un-
derstanding your individual patient’s condition—their day-to-day life,
activities, and care management as well as therapy needs, options,
and challenges—should be the starting point of designing a potential
DTx solution. What problem(s) are you trying to solve for your pa-
tient? Learning to interpret your patient’s frustrations and potential
unmet needs (as well as t hose for every other stakeholder in their

ecosystem) and converting them into a neatly defined job to be done
is a critical skill. It is not an easy task, but t hese questions are a good
starting point.
What kind of problem are you trying to solve and in what specific circum-
stance are you trying to solve it?
Fabien: The long-term side effects of years of uncontrolled diabetes have
started to manifest as oral health problems and, more recently, diabetic
neuropathy—a type of nerve damage, according to my physician. Coming
back to more acceptable glycemic levels is my top priority. In fact, it has always
been. I must be successful at managing my diet, exercise, regularly checking
my blood sugars, and giving my insulin while maintaining an active life-
style [remember he is only 25 years old] without having access to solu-
tions that have been proven to be effective for me or other patients, such as
the pump.26
From this extract of our discussion with Fabien, his glycemic variabil-
ity is clearly a short-term problem he is trying to solve. But if you w ill
be having this discussion in the f uture with a patient community, more
patient-provided specifics w ill be needed to better understand the con-
text and the reasons why, when, and how it is easier (or not) to man-
age their blood sugar.
Is t here an important problem that lacks an adequate solution?

Fabien: The main problem for me is the relationship with physicians. They
know well the theory, the disease evolution and manifestation. I even had
endocrinologists who themselves had diabetes, but each patient is differ
ent. I know myself. When I have hypos [low blood glucose], I know the
best ways to get my blood sugar back up quickly without getting hypergly-
cemia. Many times I can predict my glycemic level without testing myself.
However, my physicians are not listening to me enough. My day-to-day ex-
periences are diff erent. I can’t stick to a routine and have the same glycemic
intake every day, or the same level of physical activity and be in the same
emotional state without getting sick.

How are you getting the job done now? What are your compensating behav
iors in the absence of an optimal solution?
Fabien: I need to test myself more, and I find my ways of d oing it by creat-
ing my own habits. For example, I have associated each glycemic test with a
specific moment of my day so I d on’t forget. My blood sugar reader is, for
example, always near my alarm clock, so it’s the first t hing I do e very morn-
ing before snoozing the alarm. I also tend to have a tea break around four
p.m. While the coffee gets ready, I do a test, etc.
For Fabien, these are critical moments of the day where he is constantly
reminded of the need to take a test or simply take care of himself. Un-
derstanding these moments, which are unique to each patient, is an
opportunity to send the right notification or nudge at the right time—
which we w ill explore in more detail later.
Despite established routines and reminders to comply with diabetes
management protocols, missed steps are a common reality when living
with diabetes. This is where compensatory behaviors come in. Compen-
satory behaviors are actions patients take to make up for something
they should have done—behaviors meant to compensate for something
else, a workaround. Fabien knows he is not g oing to test himself as much
as his physician has requested, so he focuses on his food intake and has
learned over time how to listen to his body—which is not a perfect solu-
tion. Patients with long-standing diabetes may start to lose some of the
warning signs of low blood sugar (hypoglycemia) or lose the ability to
predict when they are having a hypoglycemic episode, which in turn can
result in a more severe episode. Therefore, solutions that provide action-
able blood glucose readings without Fabien consciously having to check
would be welcome and eliminate the guesswork.
What functional, social, and emotional criteria do you use to evaluate what
is right for you? Conversely, what functional, social, or emotional f actors
would take a potential solution “off the table” for you?
Fabien: I was initially against the insulin pump. Then I considered what I
would gain and found that it promised much more flexibility for meals (partly

emotional and partly social criteria), and I was able to reduce the items I
needed to carry around (functional). T hese advantages outweighed the chal-
lenges that would come with it, such as the constant reminder of being sick
with diabetes (emotional) or how o thers would perceive me if I had this de-
vice on me all the time (social). A pump would have also been “off the table,”
as you said, if the potential pain during the initial set-up or during daily
usage, or the potential risks such as the risk of infection at the infusion site,
outweighed the overall benefits.
How is “value” measured from your perspective?

Fabien: Value for me, is simply the ratio of benefit and loss for any solution,
including the financial aspects and potential co-payment. Not to mention,
if you add the patient’s willingness to be at the center of his pathology, a
convinced patient would probably benefit more from any solution.
Fabien is looking at value in the form of a yes-no decision when accept-

ing a solution, such as the pump in this example. The value is also re-
lated to the notion of success for him, that is, how he defines success
and how the value of a solution throughout the evolution of his condi-
tion, in one way or the other, w ill help him feel successful.
When conducting such interviews with patients like Fabien, t here
are important shortcuts and pitfalls to avoid. In our experience of us-
ing the JTBD approach ourselves, we strongly feel that it is equally
important to listen to what a patient says and observe what he does.
Trusting action over statements tends to be a reliable approach to iden-
tifying the underlying needs, which is often the limitation of tradi-
tional survey-based market research.
We have also realized through patient interviews that it’s important
to not fall victim to the curse of knowledge in order to avoid cognitive
bias. If you know that you are going to interview “Fabien, type 1 dia-
betes, 25 years old,” then t here may be preconceived notions and ex-
pectations about his challenges and struggles. For this reason, we
should focus on the circumstances of the stakeholder rather than their
demographic profile. We should look deeper into patterns, themes, and

insights to identify the root causes that are driving their needs to pri-
oritize their JTBD.
As we have seen with Fabien, two jobs can lead to diff erent solutions,
so the question of which to prioritize is crucial. When there is more
than one possible JTBD, consider their importance from the stakehold-
er’s perspective as well as the stakeholder’s satisfaction (or lack thereof)
with existing solutions. The larger the gap between the two (e.g., impor
tant JTBD but unsatisfied with the status quo), the clearer the target
area of a DTx platform. Based on this assessment, targets could be de-
veloping a new feature, a new functionality, a new way to use the power
of data science, or some other entirely novel solution. In the context of
the discussion with Fabien, the JTBD approach provided key insights:
1. He has a clear idea of steps needed to manage his condition.

Hence, the importance of the first job definition suggested
earlier was not high. Solutions developed to tackle this JTBD
may not find Fabien as a user.
2. Achieving the targets of the second job definition is a priority
for Fabien. At the same time, current solutions do not satisfy
his requirements. Hence, a user-identified target for DTx
development or improvement may have emerged.
To conclude, JTBD is one of many approaches available for uncov-

ering the key factors that drive an individual’s health behaviors. Re-
gardless of the approach used, it is critical for DTx organizations to
perform such analyses as they craft new functionalities to perform
better than existing alternatives and at the same time address or lever-
age current compensating behaviors. Comprehensively considering
these points w ill realize a DTx that fulfill users’ functional, social, and
emotional needs in a true customer-centric approach.
Addressing Pain Points and Delivering Value to Stakeholders

Now that w e’ve considered the importance of understanding the
patient and provider perspective when developing DTx platforms,

it’s time to think about the aforementioned objective of delivering
stakeholder value. Finding a winning business idea entails under-
standing the pain points of the industry you are working with and,
more importantly, of your stakeholders. In health care, this typically
means focusing on meeting the needs of and overcoming the chal-
lenges confronted by patients, caregivers, and physicians. Established
and aspiring DTx innovators can approach this process by systemati-
cally analyzing and empathizing with their potential future users,
understanding the pain points they experience at every stage of their
journey, and figuring out how a DTx solution can help them overcome
those barriers and unlock utility or new value (figure 2.2).
Fabien’s ideal end-to-end patient journey spans from seeing his
specialist, to being offered a new DTx that helps him manage his dia-
betes, to ultimately a “digital exit,” which is achieved when a patient
may not need the DTx any longer. This journey can be visually under-
stood by applying W. Chan Kim and Renée Mauborgne’s concept of the
Buyer Utility Map.27 According to this concept, Fabien may experi-
ence different emotions and pain points during his patient journey.
T hese emotions and pain points can be organized into five categories:
productivity, simplicity, convenience, risk reduction, and fun and im-
age. T
hese categories—also called utility levers—help user experi-
ence researchers (a role DTx innovators should consider incorporat-
ing into their teams) label and group pain points and obstacles in a
more systematic and granular way, instead of simply asking a patient
“What are the challenges you face in this journey?” and relying on
his spontaneous response at that moment.
Applying a systematic approach to the pain point discovery process
helps us stay focused on solving the real problem at hand while elim-
inating the potential myopia of pure novelty, which all too often leads
to innovators having that solution still looking for a problem. Rather,
DTx entrepreneurs should strive to create demand-driven products in-
stead of supply-driven ones. A systematic analysis of the user journey
is imperative in therapeutics in general, but especially so in digital

Ho_Meds_int_3pgs.indd 65
Stakeholder journey
Interaction
with Health Patient DTx Usage DTx Usage “Digital”
Awareness Care Prescription Onboarding (First 90 days) (>90 days) Exit
Professional
Stakeholder – Anything that has to do with helping the stakeholder
Productivity
Productivity spend less time, effort, or money (increasing efficiency)
– Anything that eliminates or minimizes complexity or
Simplicity mental hassle from the stakeholder's point of view
– Anything that creates exceptional value in
Convenience terms of availability anytime, anywhere
Utility levers
Risk – Anything that reduces stakeholder physical,
reduction emotional, or financial risk
Fun and – Anything that creates exceptional utility in terms of tangible
image Lack of immediate results or intangible aesthetic look, feel, attitude, or style
leads to discouragement and
Legend: pain points to be defined and addressed lack of motivation, which in
turn leads to nonadherence
Figure 2.2. Pulling the right levers to deliver utility. Source: Adapted from the Buyer Utility Map in Kim and Mauborgne, Blue
Ocean Strategy; for illustration purposes only
14/06/23 12:45 PM
therapeutics, b ecause it provides a s imple yet powerful framework to
test the desirability of a therapeutic while taking user centeredness
into consideration.
Once the pressure points are defined, the solution can be framed
around the findings and address each point one by one, as organized
by the utility levers. Using this framework in the postmortem analy
sis of Exubera, Pfizer’s rapid-acting inhaled insulin formula, we can
begin to appreciate why it failed to unlock exceptional utility for Fa-
bien and other diabetes patients. From a breakthrough novelty stand-
point, Exubera seemed to hold g reat promise because it potentially
meant that people with diabetes could move away from the inconve
nience and stigma of needle injections. Nevertheless, the shortcom-
ings were evident from a utility point of view.
• Stakeholder productivity. Despite the cleverness of the under

lying technology, the size of the inhaler was too big and cum-
bersome to handle. Even for a young person like Fabien,
holding it properly posed a challenge. Apart from that, using
the product was a time-consuming procedure since it required
inserting a series of blisters and activating the pump before the
inhaler was ready to be used. Of note, this process took several
minutes in contrast to insulin injections, which typically take
only a few seconds to administer.
• Simplicity. Pfizer underestimated the complexity of using their
product, which required selecting a specific insulin dose—a
difficult task and major ask for the patient and doctor. Most
patients like Fabien would have needed extra education on how
to measure and select a dose in order to use the product, which
in turn would have required that physicians and nurses spend
extra time to educate their patients.
• Convenience. The issue of convenience was one of the main
reasons for this product’s failure—it was not designed with
ease-of-use considerations. Its size made it a challenge to carry
around and its usage inconvenient.

• Risk reduction. The product was priced at a premium, three to
four times the average daily cost of injected insulin for a type 1
diabetics.28 As such, it created a financial risk instead of reduc-
ing it. To address this risk, most patients were keen to test the
device first before committing to the higher expenditure.
Pfizer, however, did not make such an option available. Apart
from that, physicians w ere concerned that it could lead to the
development of lung disease in users. In the absence of a
long-term clinical trial informing about Exubera’s safety and
efficacy, it was not recommended to children, smokers, or
patients with lung diseases, such as asthma or chronic obstruc-
tive pulmonary disease.29
• Fun and image. This is a utility lever that is often overlooked in
the health care industry b ecause of the dominant focus on
functionality, yet for some products the emotional proposition
is as important as their effectiveness, particularly when it
comes to lifestyle integration. In Exubera’s case, as previously
stated, the product was not discrete, which could lead to
unwelcome attention in groups or public. Furthermore, it
appears that the inhaler’s designers had not given much
thought to what would have made Fabien want to show off his
new gadget with style and accessories.
Outside of the diabetes arena, a striking example of successfully ana-

lyzing and addressing the fun and image lever is in the pediatrics men-
tal disorders field. Akili’s EndeavorRx, an FDA-cleared DTx for children
with attention deficit hyperactivity disorder (ADHD), is delivered in the
form of a video game and its young users are not considered patients
or customers by Akili, but “players” who steer a craft through obstacle
courses to collect targets (figure 2.3). Gamification, or the use of game
design elements, is a unique proposition that can be integrated into DTx
to reimagine treatment modalities and boost patient engagement. Play-
ing a computer game to treat a disorder—does this sound too futuristic,
or is a reimagination of care delivery in order?

Figure 2.3. Akili’s EndeavorRx video game interface. Source: Photo by Akili Inc., 2021
Gamification is a promising way to improve patient engagement and

disease self-management. By deploying behavioral strategies to unob-
trusively motivate users to complete actions, t hese games may help
users stick to their treatment plans more effectively than traditional
approaches. That said, gamifying processes requires a focused study
and understanding of the solution or product’s end users. In Akili’s
case, they are children and their caregivers.
Applying the concept of utility levers when developing a new DTx
product is essential to helping organizations avoid the risk of their of-
fering not gaining traction or failing to be adopted, despite any nov-
elty. Companies that understand the importance of utility levers and
design solutions based on this understanding can significantly derisk
their operations in solving the right problem and unlocking practice-
changing utility to patients, caregivers, and physicians.
Understanding the Buying Decision Process

fter r unning a diagnostic of the ecosystem, understanding the jobs
A
that stakeholders are trying to get done, and figuring out the levers

that need to be pulled to deliver exceptional utility, t here is one more
important consideration that needs to be grappled with: that of un-
derstanding the chain of potential buyers for a DTx and knowing where
in that chain to focus on. The concept of a “chain of buyers” was in-
troduced by Kim and Mauborgne of Blue Ocean Strategy in 2002.30 It’s
a simple yet insightful tool that helps entrepreneurs broaden their vi-
sion and narrow their focus at the same time.
The purchase of any product or serv ice involves a chain of people
who are directly or indirectly involved in the buying decision. That in-
cludes the users of the solution, the purchasers who pay for it, and
increasingly often, the influencers whose opinions can make a differ-
ence. In Europe, for example, where health serv ices are typically free
or near-free at the point of care u
nder a system of universal health cov-
erage, the chain of buyers involved in the decision-making process for
starting insulin pump therapy for a patient is strongly influenced by
the opinions of endocrinologists and health care providers directly in-
volved in the care of the person with diabetes, such as diabetes nurse
educators and dietitians. T hese professionals are important buyers
because they assist patients in analyzing clinical and economic risks,
implications, and benefits of the therapy, and patients, mostly, value
and trust the opinion of their health care providers. W hether or not a
health care provider recommends such a product to the patient is, in
turn, influenced by their knowledge of and familiarity with the prod-
uct, past experience (both positive and negative), and additional f actors
(such as reliability or responsiveness of technical support, etc.). In con-
trast, in self-pay markets such as India, the decision-making system
for the same therapy is strongly influenced by patients themselves and
their families, who generally want to understand the costs of the
pumps, supplies, and so on, as well as treatment options presented to
them by their health care provider, and compare them with the alter-
natives of nearby hospitals and providers.
Focusing on and understanding the chain of buyers who might
influence a buying decision should be one of a DTx innovator’s top

priorities as it is an important step in the ecosystem diagnosis and
in designing value propositions that address stakeholders’ jobs to
be done.
Case Study: Medtronic

Medtronic, a leader in medical technology known for its focus on dia-
betes management, saw in 2011 what the company believed was an un-
tapped market in India—pregnant w omen with gestational diabetes
mellitus (GDM). GDM is a condition in which blood sugar levels be-
come elevated during pregnancy, usually from 24 weeks of gestation
onward. While GDM is typically a short-term condition that lasts only
for the duration of the pregnancy and shortly thereafter, it increases
the risk of type 2 diabetes, a long-term condition. Additionally, GDM
can negatively affect healthy development of the baby.31 India has an
especially high rate of GDM, with prevalence ranging from 3.8% to
17.9% of all pregnancies depending on the part of the country.32 Preg-
nant women whose GDM is inadequately controlled through dietary
measures need to move on to insulin injections or start using insulin
pumps.
An insulin pump is an insulin delivery device designed to mimic the
way a human pancreas works by delivering small, precise doses of
short-acting insulin continuously and other variable amounts of insu-
lin when a meal is eaten. However, an insulin pump in India costs
between US$2,500 to US$6,500, amounts which are prohibitive for
most Indian w omen with the need for short-term diabetes control.33
The cost of the device could effectively be a barrier to treatment for a
substantial segment of the population.
Medtronic recognized the underserved market for its insulin pumps
in India, where there were about 435,000 w omen (420,000 pregnant
women with GDM as well as 15,000 w omen with type 1 or 2 diabetes
prior to the pregnancy) who w
ere eligible for a pump but w
ere not using
one. To address this gap, Medtronic decided to shift its perspective
from the influencers (traditionally physicians and endocrinologists)

T able 2.2. Influencer and user concerns regarding insulin pumps for
gestational diabetes mellitus (GDM)
Influencers (physicians) Users (patients)
• Lack of knowledge, familiarity, education, and • Low awareness of the availability of such
training on insulin pumps for GDM, leading to a technology.
limited awareness and knowledge among • Poor self-testing.
physicians. • Affordability, high upfront cost, and
• Lack of experience and conviction that insulin recurring cost of consumables (e.g.,
pumps are appropriate for managing GDM. infusion sets and reservoirs/cartridges).
• Time-consuming education and hand-holding for There is also concern about what
patients during the initial introduction to insulin happens to the insulin pump after GDM.
pump therapy. • Fear of needles/cannula going under
• Relatively high upfront and recurring costs of the skin.
insulin pumps to patients. • The dislike of having a device attached at
• Patients may not turn up for the regular all times.
follow-ups that are needed in managing GDM, • Steep initial learning curve to understand
which requires close monitoring and titration of how to operate the pump and trouble-
insulin doses with each trimester. shoot in a short t ime frame.
to the users (GDM and pre–gestational diabetes patients) and focus

on understanding their needs and addressing their barriers to treat-
ment (table 2.2).
The Medtronic team quickly identified the device cost as the most
commonly stated adoption barrier by not only patients but also by phy-
sicians who w ere reluctant to suggest what was seen as an expensive
therapy to their patients. The company thus decided to develop an
affordable solution for Indian w omen with GDM and at the same time
work to convince their physicians of the value of insulin pumps for
this condition. A fter experimenting with several business models,
Medtronic settled on rolling out a rental program. For about US$150
per month, GDM patients could rent an insulin pump instead of buy-
ing one. The company started to test how appealing an insulin pump
rental program would be for this group of potential users who w ere re-
sistant to pay the full cost of the pump upfront. Would they be willing
to rent used pumps in the first place? Was the price right?
A shift in the target segment from the influencers to the users
revealed a series of unknowns and assumptions that needed to be

reevaluated. Medtronic had to not only test how appealing their con-
cept would be to the patients and to the physicians but also assess the
overall feasibility of the program from an operational and financial
perspective. Would the physicians be willing to prescribe insulin
pumps for GDM patients? How intensive would patient training and
education need to be? Could Medtronic develop a system to track
and distribute the pumps broadly, at scale? What about payment col-
lection? Would the number of GDM patients per hospital or clinic be
sufficient to serv ice this program efficiently? The questions were nu-
merous but vital toward establishing the feasibility of this approach.
The two distribution models, purchasing and rental, w ere offered
concurrently. During the pilot program, Medtronic observed that four
times more women34 who rented the insulin pump would ultimately
buy their own pump as compared to the women who were not offered
the rental program. With this promising result, Medtronic decided to
go one step further and integrate this innovative “try before you buy”
model into its core business model for servicing type 1 and type 2 di-
abetes patients. Shifting its focus from the influencers to the users and
purchasers allowed Medtronic to not only unlock a new target seg-
ment, but also markedly increase the size of its business in India.
The “try before you buy” model, which convinced diabetes patients
to sign up for the pump therapy with a relatively small rental fee ad-
dressed both the issue of patient affordability as well as the physicians’
concerns about suggesting expensive treatment options to their pa-
tients. It also achieved a rare feat in health care: giving users an op-
portunity to try a drug or a medical device before making a long-term
commitment. This pilot program has also convinced more physicians
to get on board with recommending insulin pumps by increasing the
physicians’ knowledge and experience with the device.
Testing the assumptions that came with this shift of focus in the
chain of buyers was critical to the success of Medtronic’s program. It’s
a powerful example of how to approach a situation where technology
alone is not enough to solve a problem, especially in health care.

Medtronic had to question the practicality of its standard commercial
model in a different market such as India, invest effort and forward
thinking, and be willing to apply what it learned to revamp their busi-
ness model. In this particular case, the right strategy led to unex-
pected success that surpassed initial presumptions.
A shift of focus in the chain of buyers is not necessarily a shift from
a business-to-business to a business-to-consumer operating model, or
to a potentially more complex model. It’s also not about new brand po-
sitioning. Rather, as the Medtronic example demonstrates, it is about
making changes to your business model and unlocking new user re-
search capabilities that you may need to develop or acquire. It also re-
quires addressing stakeholder jobs-to-be-done and potential pain points
with which they may be faced, as discussed e arlier.
Conclusion
In this chapter we took a closer look at traditional diabetes manage-
ment to demonstrate the critical importance of assessing stakeholder
needs and its impact on the medical and commercial outcomes of a new
health care endeavor such as DTx. Commercial use cases demonstrated
the different ends of the spectrum that are achievable as a result of
addressing t hese needs. DTx organizations should build on these learn-
ings to successfully develop solutions that can turn invention into in-
novation, catalyzed by optimally serving the stakeholders involved.
We believe indeed that the most successful organizations are t hose
that solve problems at the intersection of patient and physician jobs-
to-be-done and unmet needs, with the solutions also meeting the re-
quirements of stakeholders further downstream. Visionary founders
and developers ask questions, listen, and study carefully the challenges
key stakeholders face to find and define their intersection because they
know that delivering value to both patients and physicians is the way
to develop a solution that w ill be a dopted. The patient-physician
perspectives outlined in this chapter serve as examples of experiences

on the ground, and t here are many more stakeholders involved in
the health care and DTx ecosystem whose pain points need to be
identified and addressed. DTx developers and entrepreneurs should
therefore be prepared to build acceptance for the changes that these
new technologies w ill inevitably cause in the broader ecosystem, in
addition to developing new technologies. As one of our contributors
reminds us,
We always start with the problems, the inefficiencies we want to address in

health care. How we may address these problems may evolve over time and
require a few pivots because of our continuous learnings, because our partners
may be diff erent, or because of the market-to-market specificities. Regardless
of the context, though, for us it’s about moving from a transactional to a
programmatic way of delivering care and doing so in a more integrated
manner to ultimately deliver tangible and proven outcomes. —A zran
Osman-R ani, CEO of Naluri, aspiring DTx company based out of Malaysia
that addresses physical and m ental health holistically through profes-
sional, confidential health coaching.35
It’s worth repeating: a problem well-defined is a problem half-solved.
Notes
1. Martino, Hollis, and Teichert, “Pharma’s Biggest Flops.”
2. Gallagher, “GP Appointments Should Be at Least 15 Minutes in Future.”
3. Wientraub, “Pfizer’s Exubera Flop.”
4. Chen and Wong, “Singapore Beats Hong Kong in Health Efficiency.”
5. Singapore Ministry of Health, “National Health Survey 2010”; National
Medical Research Council, Diabetes Taskforce Report.
6. Institute of M ental Health, “Singapore Residents Show a High Recognition
of Diabetes.”
7. World Health Organization, “Diabetes: Key Facts.”
8. Toh and Wong, “The Big Read: Apathy, Complacency—the Worst Enemies in
Singapore’s War against Diabetes.”
9. King et al., “Perceptions of Adolescent Patients of the ‘Lived Experience’ of
Type 1 Diabetes.”
10. American Diabetes Association, “Insulin Myths and Facts.”
11. Davis, Burgen, and Chen, “Out-of-Pocket Costs for Patients with Type 2
Diabetes Mellitus.”

12. Harris et al., “Factors Influencing Attendance at Structured Education for
Type 1 Diabetes in South London”; McIntyre, “Dose Adjustment for Normal Eating:
A Role for the Expert Patient?”
13. The English Index of Multiple Deprivation combines information from
seven domains to produce an overall relative measure of deprivation. The domains
are income, employment, education, skills and training, health and disability,
crime, barriers to housing serv ices, and living environment.
14. Harris et al., “Characterization of Adults with Type 1 Diabetes Not Attend-
ing Self-Management Education Courses.”
15. Brown and Bussell, “Medication Adherence: WHO Cares?”
16. Peyrot et al., “Psychosocial Problems and Barriers to Improved Diabetes
Management.”
17. World Health Organization, Adherence to Long-Term Therapies: Evidence for
Action.
18. Riddle, Rosenstock, and Gerich, “The Treat-to-Target Trial: Randomized
Addition of Glargine or Human NPH Insulin to Oral Therapy of Type 2 Diabetic
Patients.”
19. Kennedy-Martin, Boye, and Peng, “Cost of Medication Adherence and
Persistence in Type 2 Diabetes Mellitus.”
20. Egede et al., “Medication Nonadherence in Diabetes: Longitudinal Effects
on Costs and Potential Cost Savings from Improvement.”
21. Michie et al., “Evaluating the Effectiveness of Behavior Change Techniques
in Health-Related Behavior.”
22. Hood et al., “Effective Strategies for Encouraging Behavior Change in
People with Diabetes.”
23. American Diabetes Association, “The Cost of Diabetes.”
24. Christensen et al., “Know Your Customers’ ‘Jobs to be Done.’ ”
25. In this context, by “outcomes” we are referring to short-term prog ress or
wins rather than to the clinical outcomes of managing a chronic disease.
26. As per guidance from the National Institute for Health and Care Excel-
lence’s Continuous Subcutaneous Insulin Infusion for the Treatment of Diabetes
Mellitus, “insulin pump therapy is recommended as a treatment option for
adults and children 12 years and older with type 1 diabetes mellitus provided that:
[1] attempts to achieve target haemoglobin A1c (HbA1c) levels with multiple daily
injections (MDIs) result in the person experiencing disabling hypoglycaemia . . .
defined as the repeated and unpredictable occurrence of hypoglycaemia that
results in persistent anxiety about recurrence and is associated with a significant
adverse effect on quality of life, or [2] HbA1c levels have remained high (that is, at
8.5% [69 mmol/mol] or above) on multiple daily injections therapy (including, if
appropriate, the use of long-acting insulin analogues) despite a high level of care”
(p. 4).
27. Kim and Mauborgne, Blue Ocean Strategy.
28. Japsen, “Inhaled Insulin’s Cost May Take Breath Away.”
29. Barclay, “Exubera Approved Despite Initial Lung Function Concerns.”

30. Kim and Mauborgne, Blue Ocean Strategy.
31. Wei, “Nurture or Nature? Research Project Reveals Answers.”
32. Chudasama et al., “Magnitude of Gestational Diabetes Mellitus.”
33. Kesavadev et al., “Use of Insulin Pumps in India: Suggested Guidelines
Based on Experience and Cultural Differences.”
34. Women with preexisting diabetes and not GDM.
35. From an interview with the authors, June 24, 2020.

• Chapter 3
Addressing Unmet Clinical Needs
Brian Harris is a board-certified music therapist and one of only 350

people in the world to hold the distinction of Neurologic M
usic Ther-
apy Fellow. Seeing firsthand the impact of neurologic m
1
usic therapy
(a research-based system of 20 standardized clinical techniques for
sensorimotor training, speech and language training, and cognitive
training) on his patients at the Spaulding Rehabilitation Hospital in
Boston who had suffered such neurologic injuries as stroke, brain in-
jury, or Parkinson’s disease, Harris became convinced of its therapeu-
tic value and began searching for a way to expand its impact beyond
the rehabilitation hospital. That’s how MedRhythms was born in 2015.
A neurorehabilitation company Harris cofounded with business entre-
preneur Owen McCarthy, MedRhythms focuses on developing and
bringing to market DTx solutions that use sensors, music, and soft-
ware to digitize therapeutic interventions based on the neuroscience
research of how m usic is processed by the brain.
Music has been shown to activate the areas of the brain responsi-
ble for movement, language, attention, memory, executive function,
and emotion.2 Thus, the first question MedRhythms had to answer at
77

the start of its DTx endeavor was which of these areas they were to
focus on:
We decided to initially focus on walking because of the strong evidence and

research in this field. There are over 50 clinical research studies demonstrat-
ing improvements in movement when using Rhythmic Auditory Stimulation
(RAS). One of those studies showed that each 0.1 meter per second increase in
walking speed [as a result of RAS] was correlated to a 7% decrease in fall
risk.3 It was also important to have the ability to measure improvements, such
as walking speed and walking quality. The overall objective was functional
improvement.
Regarding the indication, t here was also strong evidence with trials
demonstrating improved walking when RAS was delivered for a rehabilitation
after stroke, and in multiple sclerosis (MS), cerebral palsy, and Parkinson’s
disease. We narrowed the therapeutic areas down to MS, for which there was
already a drug on the market, and stroke, where there was a high unmet need
with literally no standard-of-care and no solution. In the US alone, t here are
about 3.5 million chronic stroke patients with walking deficits who have been
told they would not be able to walk better again. The cost to the health and
social care system is substantial.
—Owen McCarthy, president and cofounder of MedRhythms, member of
the Digital Therapeutics Alliance Board of Directors4
For any pharmacological, medical, or health technology innovation,

choosing a therapeutic area (TA) to focus on is ultimately about ad-
dressing unmet clinical needs, a process that begins with a thorough
clinical need assessment. For DTx innovators, this means defining and
prioritizing the gaps and deficiencies in the current delivery and prac-
tice of health care that a potential DTx solution could address. T here
are many diff erent ways to go about this, but we suggest starting with
five fundamental questions that w ill help you identify the right TA to
focus on and guide your reflective process:
1. What are the conditions driving health care utilization?

2. Where is the demand outgrowing health care capacity?

3. What are some underrecognized and undertreated conditions
for which a DTx could provide a solution?
4. Who can benefit from behavioral change interventions?
5. What treatment pathways can be improved with a more per-
sonalized approach?
Let’s do a deep dive into each of t hese questions.
What Are the Conditions Driving Health Care Utilization?

In a 2008 study exploring the relationship between multimorbidity
(concurrent diseases and combinations of diseases a person may have)
and use of general practice serv ices in the Dutch population for t hose
55 years and older, researchers wanted to understand how multimor-
bidity, which is common among older p eople, affects the demand for
health serv ices. One of the variables they examined, which is seen as
a proxy for health care resource utilization, was the number of con-
tacts each of the patients had with a general practice per year (table 3.1).
A contact was defined as a face-to-face consultation, a phone contact,
Table 3.1. Total number of contacts with general practice for people with one
or more chronic diseases (2014)
Number of chronic diseases
Disease 1 2 3 4 ≥5
Heart failure 16.7 22.0 26.6 30.0 36.7

Diabetes 14.4 18.5 23.3 26.6 32.3
Chronic obstructive pulmonary disease 13.6 17.2 22.4 25.3 32.1
Stroke 12.9 17.7 21.8 25.1 29.4
Anxiety disorder 12.8 17.6 22.1 25.6 —
Depression 12.7 17.7 23.2 26.8 32.3
Coronary heart disease 11.8 17.1 21.8 25.7 31.7
Cancer 11.5 16.8 21.4 23.1 32.0
Osteoarthritis 10.8 16.2 20.7 23.4 28.4
Chronic back or neck disorder 9.4 14.7 19.2 23.8 27.9
Source: Van Oostrom et al., “Multimorbidity of Chronic Diseases and Health Care Utilization
in General Practice.”
Addressing Unmet Clinical Needs 79

or a home visit initiated by or on behalf of each patient with one or
more chronic diseases.
The mean of the total number of contacts per year was 6.1 for indi-
viduals without any chronic diseases (people who can be categorized
as “healthy individuals” for simplicity purposes), 11.7 for individuals
with a single chronic disease, and 18.3 for individuals with multiple
diseases. T hese statistics confirm what most of us intuitively know:
that patients with more chronic conditions tend to need and use more
health serv ices. It is interesting to note that in the study, heart fail-
ure, diabetes, chronic obstructive pulmonary disease, stroke, anxiety
disorder, and depression—in this order—drove the greatest number
of contacts for p eople who had one or more of t hose conditions. As the
researchers noted, a higher number of chronic diseases was associated
with more contacts but also with more written prescriptions, and more
referrals to specialized care.5
The relationship between health conditions that result in high use
of health care resources and health system sustainability and efficiency
(or inefficiency) is complex and nonlinear. While a discussion of this
correlation and its impact on health economics in general is beyond
the scope of this book, we would venture that resource-intensive health
conditions are a good starting point for DTx innovators. They provide
a sound rationale for exploring if and how a DTx can benefit both the
patients with these conditions and the broader system by delivering
health improvements in a more cost-efficient manner than current
treatments and their associated resource requirements do.
It is important to keep in mind that while the conditions mentioned
in the study w ere identified as leading to frequent health care resource
utilization in the Dutch health care system, in other countries and ge-
ographies the makeup of the costliest diseases may be different. In
this broader health care context, however, t hese figures provide impor
tant insight for DTx entrepreneurs to consider where they may have
impactful opportunities to develop and implement their solutions.

Where Is the Demand Outgrowing Health Care Capacity?
Resource-intensive diseases are associated with an increased demand
for health services that in many places are outgrowing the system’s ca-
pacity. Globally, nurses and midwives account for nearly 50% of the
health workforce and are the backbone of many health care systems,6
yet they routinely spend precious time performing administrative
tasks instead of providing direct care to patients. As a consequence,
many of them feel that their time is mismanaged and could be better
spent if they were instead allowed to focus on managing, supporting,
and educating patients. Taking nurses away from direct patient care
and forcing excessive admin work upon them is particularly inadequate
when we consider that many health care systems struggle with a short-
age of nurses—1,000 p eople can be attended to by 11.7 nurses in the
United States, 9 in countries that are members of the Organisation for
Economic Co-operation and Development (OECD), and only 2.7 and
1.7 in China and India, respectively.7
The situation with regard to specialists such as ophthalmologists,
neurologists, and psychiatrists is similar. The shortage of specialists
makes apparent the inequality of access to health care between high-,
middle-, and low-income countries. For example, for every 1 million
people in the developed markets, there are 76.2 ophthalmologists. In
the developing markets this number drops to 3.7 ophthalmologists per
1 million people.8 With a growing aging population worldwide, the
problem of health care resource shortages across therapy areas is only
expected to deepen.
W hether you are in a developed or developing market, there is a high
chance that you are g oing to quickly find areas along the care contin-
uum and within specialties where the demand for health care serv ices
is outgrowing health care system capabilities. Examining in-depth
these areas and w hether your potential DTx solution can help allevi-
ate t hose pressures, could constitute an important element of your
value proposition.

What Are Some Underrecognized and Undertreated Conditions
for which a DTx Technology Could Provide a Solution?
here are several thousand known diseases that affect h
T umans, but
only about 500 of them have treatments approved by the US Food and
Drug Administration (FDA).9 In addition, for some of t hese diseases,
medication can only help control the symptoms and cannot cure the
underlying cause. There are more than 6,800 of t hese diseases affect-
ing an estimated 25 million to 30 million Americans10 and are typically
not financially attractive targets for developing novel pharmacological
solutions for because of the small markets each represents individually.
While not v iable as business opportunities to pharmaceutical compa-
nies mainly due to the lengthy, expensive, and high-risk drug develop-
ment process, rare diseases still account for a huge proportion of the
health burden both to patients and their families, and to the health
care system at large.11 They could be a good target for DTx development
due to the potential of markedly shorter times to market.
Among the more prevalent conditions for which t here is currently
no effective treatment, but could be ripe for digital disruption, are Alz-
heimer’s disease and Parkinson’s disease. Billions of dollars have been
invested in research into potential cures for both, but efforts to develop
drugs that effectively treat memory loss and tremor (the primary
symptoms of each of t hose diseases, respectively) have so far been un-
successful. While a recent FDA approval for an Alzheimer’s drug to be
used was granted in 2021—the first of its kind in nearly 20 years—
promising drugs have repeatedly failed in phase 3 clinical studies.12
Alzheimer’s disease and other dementias contribute to human suffer-
ing at scale, and both Alzheimer’s and Parkinson’s have been listed
among the World Health Organization’s top 10 diseases that cause the
most deaths worldwide.13
Another example of underrecognized and undertreated conditions
are neurological diseases leading to cognitive impairments, such as
schizophrenia, stroke, autism, and traumatic brain injury. Among
those, stroke is the main cause of long-term disability. Up to six months

a fter suffering a stroke, two-thirds of survivors have difficulty walk-
ing, and beyond six months over 30% still cannot walk without assis-
tance.14 While individuals with stroke receive rehabilitation training,
it rarely lasts more than a few months after reaching a plateau in func-
tional recovery.15 Often due to a lack of resources for providing long-
term serv ices, this is where the science of neuroplasticity, as the basis
of an effective DTx solution, could potentially help these patients.
The ultimate goal of stroke care is to recover and maximize the brain
functions that have been lost due to cerebral damage. Brain rehabili-
tation therapy focuses on providing stimulation for neurons to pro-
mote brain plasticity—which is the ability of neural wiring to change
(e.g., abandon old and form new connections). To further bolster this
segment of treatment, digital neurotherapeutics have recently emerged.
This subcategory of DTx has the potential to specifically target health
deterioration associated with cognitive impairments by targeting the
neurological functions that can be remedied with cognitive training.
There are various technologies being developed and used in DTx that
can deliver “repetitive, goal-oriented tasks with variability while
providing performance feedback.”16 To put it more simply, think of
the DTx solutions we have discussed so far and their targeted stimula-
tion strategies—for example, Akili Interactive using a video game
(see chapter 2) and MedRhythms using m usic—as a “gym for the brain”
that strengthens and expands patients’ cognitive capabilities.
Identifying an unaddressed gap to target your DTx requires careful
study and understanding of patients’ lived experiences with under-
recognized and undertreated conditions. For t hose with conditions
for which t here is no existing treatment or cure, effective symptom
management is still the best option. This is exactly where DTx can be of
greatest value. We saw an example of this at the beginning of this book
in the context of Simon’s experience with pain management, where an
intelligently designed therapeutic app helped him in ways no pharma-
cological treatment was able to achieve. Pain management is also a
topic of utmost importance to the chronic pain and palliative/hospice
care communities, where demand is outgrowing health care capability

and capacity. A pain management DTx may offer an opportunity to
mimic or replicate to a certain extent the reach of effective pain care to
those who cannot avail themselves of it via conventional channels,
effectively creating a new care delivery model.
Who Can Benefit from Behavioral Change Interventions?

Following years of research, clinical studies, and practice, cognitive
behavioral therapy (CBT) has become a widely used therapeutic ap-
proach and is considered a gold-standard in psychotherapy17 “for a
range of problems, including depression, anxiety disorders, alcohol and
drug use problems, marital problems, eating disorders, and severe
mental illness.”18 Over the years, many traditional cognitive behavioral
therapists have also transitioned to digital formats, creating a digital
cognitive behavioral therapy (dCBT) category. This is the direction
taken by Omada Health, t oday one of the most highly valued digital
health companies.
In 2011, Omada Health decided to “revolutionize health care
through group-based programs for chronic disease prevention,” and
ere initially focused on diabetes.19 Specifically, the com
their efforts w
pany digitized evidence-based behavioral treatments that were tradi-
tionally conducted in clinical face-to-face settings, such as the Centers
for Disease Control and Prevention’s Diabetes Prevention Recognition
Program in the United States. Since then, Omada Health has expanded
the digital delivery of behavioral therapy to other indications, too, ex-
panding its programs to address musculoskeletal, prediabetes and
weight management, hypertension and behavioral health (i.e., focus-
ing on improving m ental health) categories.
This expansion to other conditions represents a common theme at
the core of many DTx solutions—t he possibility to leverage mecha-
nisms to drive behavioral change aimed at stabilizing, improving, or
potentially reversing a condition. Blue Note Therapeutics is also a case
in point. They aim to assist cancer patients who commonly experience
psychosocial distress, anxiety, or depression.20 As a result, Blue Note

Therapeutics’s solution incorporates cognitive behavioral stress man-
agement (CBSM), an evidence-based protocol adapted from CBT and
relaxation stress management, designed specifically for patients with
cancer. Clinical studies have indeed indicated that CBSM can improve
cancer patients’ emotional well-being, physical health, and overall
survival.21
As these real examples demonstrate, DTx platforms are seeking to
address an increasing range of diverse medical conditions, such as anx-
iety disorders, depression, m ental health, pain management, aller-
gies, rheumatoid arthritis, gastrointestinal, insomnia, and substance
abuse disorders, just to name a few. Behavioral change strategies such
as those embedded in DTx leveraging dCBT are also designed to im-
prove therapy adherence—a critical problem across most therapy ar-
eas, which has significant clinical and economic implications.
If research into the effectiveness of digital modalities in targeted
areas is still being conducted, conditions that can benefit from CBT,
and dCBT by extension, would be a good starting point for DTx inno-
vators to consider, and of potential g reat interest for patients or f uture
patients.22
What Treatment Pathways Can Be Improved with a More

Personalized Approach?
Since no two patient profiles are alike, health care needs to move be-
yond the familiar one-size-fits-all models of prevention, diagnosis, and
treatment. Following clinical trials, real-world implementation of
treatment approaches is confronted with highly diverse patient pro-
files, and the administration of treatments derived and validated us-
ing population-based strategies is too often based on trial and error,
with physicians following available guidelines and their own experi-
ence. If a particular configuration turns out to be ineffective, the phy-
sician proceeds to prescribing a second-line treatment, then a third-
line one, and so on until identifying one that works. That is a common
pattern that affects a broad spectrum of patients ranging from t hose

with hypertension to cancer. In these scenarios the same therapy may
have diff erent results in two patients with the same diagnosis. In fact,
even a single patient’s response to the same therapy may change over
the course of treatment. The shift t oward truly personalized care is
therefore a particularly important opportunity for DTx.
To try and address this health care challenge, our team at the N.1
Institute for Health at the National University of Singapore has been
pioneering N-of-1 (personalized) clinical trials to optimize treatment
outcomes at the single-patient level. In 2018, for example, a patient with
advanced prostate cancer was provided with a combination of an ex-
perimental drug and an approved cancer medication. The patient re-
sponded to treatment, demonstrated by a decreasing prostate-specific
antigen (PSA) level. However, the patient also experienced substantial
side effects that could have led to the end of treatment. We used our
proprietary artificial intelligence (AI) technology, called CURATE.AI,
to use only this patient’s drug and dosing information to construct an
individualized profile from which an optimal dose could be recommend
to the clinical team, which represented an N-of-1 strategy for patient
intervention. Following clinician approval, the drug doses were ad-
justed accordingly. Notably, CURATE.AI subsequently suggested a
50% reduction in drug dosage, which immediately led to the lowest re-
corded levels of PSA for this patient with tumor regression a fter 16
months.23 Furthermore, the patient’s subsequent dosing data were used
to continuously update the profile, which resulted in dynamically ad-
justed dosing within established limits, representing a very different
approach compared to traditional high-and fixed-dose therapy. As
such, the treatment with the experimental drug combination com-
bined with CURATE.AI dose optimization sustained treatment re-
sponse and drug tolerability, ultimately improving the patient’s qual-
ity of life as a result of lower side effects.
This case demonstrates the potential of personalizing medicine to
achieve better health outcomes, and the applicability of this field
stretches well beyond cancer. Chronic diseases, respiratory diseases,

and multiple sclerosis, to name a few, could benefit from a more per-
sonalized approach as well.24
While N-of-1 strategies offer hope to patients in terms of individu-
alizing care, the issues of scalability and financial sustainability are
often at the forefront of discussions concerning the deployment po-
tential of personalized medicine. When these issues are addressed by
the technology itself and its pathway toward implementation, mark-
edly reducing health care costs while providing equivalent or better
health outcomes can potentially be achieved simultaneously.25 Also,
from an infrastructure requirement standpoint, this case study used
a small data approach to individualize treatment that did not require
large-scale population-w ide data to inform the dosing process. With
regard to outcomes, the risks of hospital admission for adverse events
may have also been reduced. While further large-scale clinical and
health care economics studies are underway or imminent, t hese and
other considerations demonstrate the potential of harnessing scalable
personalized medicine platforms as a means to improving the cost of
care and treatment outcomes.
In summary, if DTx can provide better clinical and financial out-
comes by taking a more personalized approach to many treatment
pathways, the solution’s economic impact is often a complex question
that DTx innovators should not overlook. They w ill be required to
demonstrate both the cost and the effectiveness of their solutions.
The Opportunity for DTx

Algorithms similar to CURATE.AI form an essential part of the engine
that powers DTx. They incorporate hundreds of individual patient-
level variables that provide visibility and insight into what a fully cus-
tomized therapy for that particular patient may look like. You may re-
call our discussion in chapter 1 of the BlueStar® Insulin Adjustment
Program, which enables diabetes patients to benefit from automated
real-time adjustments to their insulin—that is a prime example of the

power of DTx to improve patients’ lives by taking the guesswork out
of medication dosing. The ability of these strategies to empower pa-
tient adherence, which can in turn impact treatment outcomes and
subsequent health system costs, cannot be overstated.
By harnessing DTx to realize N-of-1 interventions, it is possible to
individualize and optimize care at the single-patient level, using their
own data. This idea and its operationalization represent a game-
changing shift in the way we approach health care and can lead to a
genuinely personalized and scalable medicine, for better health out-
comes and potentially lower costs.
DTx platforms have the potential to accelerate personalization in
health care with customized algorithms, but also with more personal-
ized interventions including behavioral change interventions. As we
w ill see later, and continuing with diabetes as an example, patients
using a DTx may be prompted to check their blood glucose in the
mornings if the system realizes they have not been checking it regu-
larly. They may also receive continuous real-time feedback, based on
longer-term data patterns, of their own disease monitoring coupled
with external information synthesized from a wide range of sources.
Therefore, serial engagement with DTx could change the way we think
about data in health care. For example, DTx-driven acquisition of lon-
gitudinal data could catalyze major advances in the types of data
gleaned, the ease with which data are obtained, and how those data
are acquired from individual patients. This collective shift in data ac-
quisition is what will allow for DTx-centric, practice-changing ad-
vances in health care delivery.
Conclusion
The five questions outlined in this chapter are intended to guide DTx
innovators and entrepreneurs in defining and prioritizing deficiencies
in the current delivery and practice of health care, with an eye toward
identifying which of these gaps digitization and technology are uniquely
equipped to address. In this process, identifying the therapeutic areas

with greatest unmet need and understanding the potential f uture con-
sequences of not addressing t hose gaps, or not addressing them early
enough, is of critical importance. Resolving some of them may require
a prescription drug, medical device, or digital intervention, while others
may not. Figuring out these intricacies and nuances may suggest diff er
ent paths to commercialization with different partnerships and distri-
bution models (i.e., with or without a pharmaceutical partnership). It
w ill also influence the process of choosing a theoretical framework to
guide the design and implementation of the DTx.
From our experience, this latter part of the process tends to be un-
derestimated, despite the fact that a theoretical framework presents
numerous benefits for the development, evaluation, and implementa-
tion of DTx.26 For this reason, we asked Dr. Geck Hong Yeo, a scientist
trained in the application of h uman development and psychological
theories, and Matt Oon, CEO and founder of m ental digital health
company Acceset, to share their respective academic and entrepreneur-
ial perspectives on how digital platforms can serve as a bridge be-
tween evidence-based theoretical frameworks and the design of ther-
apeutic solutions (please see appendix B).
Notes
1. Academy of Neurologic Music Therapy, “Academy Affiliate Roster by Name
and Residence,” accessed November 17, 2022, https://nmtacademy.co/findannmt. The
Academy for Neurologic M usic Therapy is an organi zation whose mission is to
disseminate, advance, and protect the practice of neurologic music therapy worldwide.
2. Rossignol and Jones, “Audio-Spinal Influence in Man Studied by the H-Reflex
and Its Possible Role on Rhythmic Movements Synchronized to Sound”; Trimble
and Hesdorffer, “Music and the Brain: The Neuroscience of Music and Musical
Appreciation.”
3. Verghese et al., “Quantitative Gait Markers and Incident Fall Risk in Older
Adults.”
4. From an interview with the authors, December 8, 2020.
5. Van Oostrom et al., “Multimorbidity of Chronic Diseases and Health Care
Utilization in General Practice.”
6. World Health Organization, Global Strategic Directions for Strengthening
Nursing and Midwifery 2016–2020.
7. Organisation for Economic Co-operation and Development, Health at a
Glance 2011: OECD Indicators.

8. Resnikoff et al., “Estimated Number of Ophthalmologists Worldwide
(International Council of Ophthalmology Update): W ill We Meet the Needs?”
9. National Center for Advancing Translational Sciences, “About NCATS.”
10. National H uman Genome Research Institute, “Rare Diseases FAQ.”
11. Angelis, Tordrup, and Kanavos, “Socio-Economic Burden of Rare Diseases.”
12. Fraunhofer Institute for Algorithms and Scientific Computing SCAI,
“Preventive Treatment for Alzheimer’s and Parkinson’s Disease”; Belluck and
Robbins, “FDA Approves Alzheimer’s Drug Despite Fierce Debate Over W hether It
Works.”
13. World Health Organization, “The Top 10 Causes of Death.” Other diseases
in this ranking include coronary artery disease; stroke; lower respiratory infec-
tions; chronic obstructive pulmonary disease; trachea, bronchus, and lung cancers;
diabetes; dehydration due to diarrheal diseases; tuberculosis; and cirrhosis.
14. States, Pappas, and Salem, “Overground Physical Therapy Gait Training for
Chronic Stroke Patients with Mobility Deficits.”
15. Grefkes and Fink, “Recovery from Stroke: Current Concepts and Future
Perspectives.”
16. Choi et al., “Digital Therapeutics: Emerging New Therapy for Neurologic
Deficits a fter Stroke.”
17. David, Cristea, and Hofmann, “Why Cognitive Behavioral Therapy Is the
Current Gold Standard of Psychotherapy.”
18. American Psychological Association, “What Is Cognitive Behavioral
Therapy?”
19. Kerr, “When Omada Health Was Young: The Early Years.”
20. Adler and Page, Cancer Care for the Whole Patient; Naser et al., “Depression
and Anxiety in Patients with Cancer”; Alagizy et al., “Anxiety, Depression, and
Perceived Stress among Breast Cancer Patients.”
21. Blue Note Therapeutics, “Cancer-Related Distress.”
22. Luo et al., “A Comparison of Electronically Delivered and Face-to-Face
Cognitive Behavioural Therapies in Depressive Disorders”; Fitzsimmons-Craft
et al., “Effectiveness of a Digital Cognitive Behavior Therapy–Guided Self-Help
Intervention for Eating Disorders in College W omen.”
23. Pantuck et al., “Modulating BET Bromodomain Inhibitor ZEN-3694 and
Enzalutamide Combination Dosing in a Metastatic Prostate Cancer Patient using
CURATE.AI, an Artificial Intelligence Platform.”
24. Hult, Measuring the Potential Health Impact of Personalized Medicine; Noell,
Faner, and Agustí, “From Systems Biology to P4 Medicine.”
25. Kasztura et al., “Cost-Effectiveness of Precision Medicine.”
26. Moller et al., “Applying and Advancing Behavior Change Theories and
Techniques in the Context of a Digital Health Revolution”; Murray et al., “Evaluat-
ing Digital Health Interventions: Key Questions and Approaches”; Manta,
Patrick-L ake, and Goldsack, “Digital Measures that M atter to Patients.”

• Chapter 4
Ensuring Business Viability Early
Please note that the “Financing Your Venture” section of this chapter
was written with contributions by Benjamin Belot, partner at Kurma
Partners, a European-based venture capitalist specializing in health
care, including having two funds focused on therapeutics as well as on
diagnosis and digital health.
As you further your understanding and exploration of important as-

pects of DTx development, sooner or later you are bound to have to
tackle the question of business viability. Fully researching the condi-
tions for the potential concept or venture’s economic viability is partic-
ularly important for innovators and entrepreneurs who are trying to
translate passion into a solid business proposition.
Ensuring the business viability of any new venture—not just DTx—
always starts by mapping out the growth projections and defining
what needs to be true for the business model to reach the desired mile-
stones and targets. Concepts such as the total addressable market (TAM),
serv iceable available market (SAM), and serv iceable obtainable mar-
ket (SOM) are considered the fundamental market size metrics and
are a good starting point.1 Studying t hese concepts and backing them
91

with market research w ill validate the market potential for your con-
cept. Is there a potential market and could it be served in a way that
turns your innovation into a profitable business (e.g., economic viabil-
ity)? Is it worth your time and resources? Could it be an investment
opportunity for potential investors?
Investors w ill want to see that you have done your research and set
ambitious but realistic revenue goals (and not the biggest possible
numbers). In addition to the TAM, SAM, and SOM, there are many
tools and frameworks out t here that can help you in assessing w hether
there is a real revenue opportunity available for your venture. We are
not g oing to list them all. Instead, in this chapter we introduce one of
the simplest yet most powerful tools to use early in your journey: the
reverse income statement (figure 4.1).
The reverse income statement should affirm that you are moving
in the right direction and confirm the business viability of addressing
the problem with the tool you identified per the previous chapters. The
premise of the reverse income statement is to “envision the unknown”
via “discovery-driven planning” applied to strategy, business, and mar-
keting of a new venture.2 It starts by determining the target profit
margin or annual profits for your venture (first cell on the left); this is
typically a measure of profitability over three years or five years. This
first assumption depends on your solution and an initial business plan
that reflects your TAM, SAM, and SOM.
In an example of hypot hetical company Newco Inc., focusing on
stroke survivors and chronic pain management in the United States,
we have set an annual profit goal of $5 million by year five, which is a
rather ambitious profitability target considering that most digital
ventures focus on customer acquisition and top-line growth to drive
a company’s sales and revenues. As a result, you may want to start us-
ing the reverse income statement by determining your target annual
revenues first. Otherw ise, after defining the profitability target,
simply work backward to define annual revenues and operating costs.
For Newco Inc., we assumed a 50% gross profit margin. (For reference,
target gross profit margins hover around 80% for digital health

What resources will it take to find, identify,
convince, and support over 725 patients every month? There are over 7 million stroke
How resource intensive will these sessions be? survivors in the US. More than half of
Ho_Meds_int_3pgs.indd 93
How many headcounts are required? How many sites/ them (3.5M) have persistent walking
clinics? How scalable is the solution? deficits and do not have access to
How profitable can each revenue stream be? Total population
physical therapy or insurance
How diversified are the current revenue 1.4M coverage. We assume that 40% of
streams? How diversified should they be? Annual # of patients
these patients (1.4M) may be willing
What would it take to triple the revenues? 8,700 to consider a new method for
Stroke Revenues Penetration addressing their walking deficits.
US$13M Price per session 0.6%
ANNUAL REVENUES US$100 Estimated average cost of a physical
US$2OM Chronic Pain therapy cost, assuming completely To reach $5M in annual profits,
Revenues Average # of out of pocket/ without insurance it requires capturing over 0.6% of
US$7M sessions/ patient all stroke survivors in the US with
ANNUAL PROFITS Estimated average gait-training walking deficits who would want
15 sessions/patient to consider a new solution.
US$5M
COGS* Selling & Marketing Salesforce

OPERATING COSTS ... Marketing, business development, and
SG&A Costs*** reimbursement
US$1OM
Overhead Costs** General & Admin Clinical advisors
... Regulatory advisors
R&D Costs Administrative assistants
Engineers Management
Techs
* COGS (cost of goods sold) refers to the direct costs of producing the goods sold by a company.
** Overhead costs refer to everything except for direct labor, direct materials, and direct expenses.
*** SG&A costs are the selling, general, and administrative expenses not directly tied to making a product or performing a service.
Figure 4.1. An example of a reverse income statement for the hypothetical company Newco Inc. (for illustration purposes
only). What needs to come true for Newco to reach $5 million in annual profits by year five? Source: From the authors
14/06/23 12:45 PM
company Livongo3 and 60% for both telehealth company Teladoc4 and
medtech company Medtronic.5) Continue to work backward on the re-
verse income statement and assume Newco Inc. has two main serv ice
lines coming from stroke survivors with walking deficits (revenue
stream #1) and from persons with chronic pain (revenue stream #2).
At some point we arrive at the near f uture, where we can compare the
inferred state of your business with what is realistically achievable.
The reverse income statement forces you to make assumptions, de-
fine what success would look like, and therefore define the basic eco-
nomics of the business. Essentially you need to ask yourself, “What
needs to happen to achieve the outcomes I am (or my investor is) seek-
ing?” It could be a series of interim goals in terms of the number of
users, the number of providers, the potential uptake of a new solution,
or other factors for your venture to be v iable. Finally, link these de-
sired outcomes to what might drive them.
The more assumptions you make, the more complex your reverse
income statement w ill be. Stress test each assumption with a what-if
scenario in order to visualize how situations outside your control—
such as economic changes to the health care system, new health care
regulations, or current events (e.g., global pandemics)—might impact
your plan. We also recommend attaching corresponding activities and
milestones that you would need to reach for those assumptions to come
true. For example, if you are concerned about your ability to reach a
certain number of clinics that serve a certain number of patients in a
given year, it may help to define what would it take to secure a single
clinic in terms of resources (e.g., manpower and marketing efforts re-
quired), success ratio (i.e., for one clinic you may need to engage 10
other clinics), lead time (i.e., how long would it take you to convert a
single clinic from a prospect to a customer, gaining their willingness
to prescribe your solution), and so on. In short, have a detailed plan.
Working through the hypothetical example in figure 4.1, one would
conclude that reaching the target gross profit margin of $5 million by
year five would require serving over 700 stroke survivors with walk-
ing deficits every month, in addition to the pain management patients

we didn’t factor in for this example. Is this logistically feasible? Tech-
nically possible? Based on this analysis, reaching the target profit level
requires serving the vast majority of Newco’s total population (i.e., its
total addressable market). These are the patients who could potentially
benefit from the solution in a given market, country, or city within a
l imited period (and probably l imited financial resources). Therefore,
Newco’s concept and go-to-market plan are unlikely to be feasible.
Keep in mind that t here are other planning tools besides the reverse
income statement one can use to assess the business viability of a DTx
concept. If you are interested in learning more about them, we recom-
mend you take a look at the following:
• Pro forma operations specs, which lay out the operations needed
to produce, sell, serv ice, and deliver a serv ice to an end user—
in other words, the activities required to run the business;
• Key assumptions checklist, which aids to ensure that assump-
tions are checked and validated; and
• Milestone planning chart, which specifies the assumptions to be
tested at each project milestone.6
Our focus in the Newco example was on driving profitability, but we

understand that the success of your venture and potential investors
may not be solely defined by profitability. It’s important to consider
the competitive dynamics of your industry. In a new area like DTx,
where market education is required, there may be a need to buy into
the market and establish dominance at the expense of short run prof-
itability in order to become the market leader. Keep this in mind as
you use some of the tools mentioned in this chapter.
Financing Your Venture

At this stage of your DTx development, you probably have a good
idea of how big your company should be within the next two to three
years (e.g., number of employees, number of products, etc.), as well as
your patient and physician acquisition strategy and costs. Soon after
Ensuring Business Viability Early 95

Key Takeaways
Achieving both profitability and growth is a delicate balancing act for any business
venture, especially for a new domain such as DTx. The reverse income statement
allows you to define a desired profit or revenue target depending on where you are
in finding this balance and work backward from there. It also helps you check the
assumptions you have built into your reverse income statement and determine if
there is a revenue model and cost structure that would be optimal for delivering
this target. The reverse income statement may serve as an effective initial check to
shed light on your planned venture’s potential business viability.
determining the market potential and business viability of your solu-

tion, you w ill need to start thinking about raising capital to support
your budding company. In the startup world, this usually entails
organizing your first round of funding, onboarding investors, and de-
veloping the capacity to think not only as a CEO with a growth mind-
set, but also as a CFO looking at the financial management of the ven-
ture. Since we are talking about startups and we have covered a
diverse range of topics that drive innovation, you may also need to be
prepared to think simultaneously from many other perspectives,
ranging from operations to business development and beyond.
How to Choose the Right Investor for Your Startup

Regardless of your venture’s development stage, getting investors on
board is an important decision for any entrepreneur. It goes beyond
the financial aspects, often addressing the entrepreneur’s concerns
about the potential dilution of control too early.7 You may have heard
this before, but it’s worth saying again: ideal investors should repre-
sent a lot more than a check. Funding your startup should be focused
on how investors w ill add value to you and your team, as well as how
their investment would accelerate and derisk your growth plan. We in-
vite you to consider the following two questions when choosing the
investors:

What assets, capabilities, and resources would the investors bring in? The
investors’ networks and their ability to facilitate introductions are of-
ten the most tangible short-term benefits to a startup. Entrepreneurs
should also look for complementary expertise that investors with ex-
perience in health care, technology, or business can bring to the t able.
This relates to our previous mention that DTx should be the best of
these three worlds. So what expertise is the most critical to your ven-
ture in order to advance to the next round of funding? Does this crit-
ical expertise include clinical trial design and initiation, obtaining reg-
ulatory approvals and reimbursement, or perhaps how to form a
partnership with a larger pharmaceutical company? What expertise
do you already have in your team and what are you missing?
How well would you work with the investors? The relationship between
a startup founder and investors can be complex, but keep in mind that
both of you have the same goal at the end of the day: to see your ven-
ture grow and make an impact, both clinically and financially. The
paths to get t here may be diff erent as your ambitions and appetite for
risk may differ. Expectations and values must be aligned early on to
get the right chemistry. Investors should be considered as partners
because they share your vision and also because some are likely to join
your company’s board of directors. Choosing the right investor is akin
to choosing the next member of your leadership team.
Similar to how you have a few rounds of discussions before select-
ing members of your C-suite, it is important to do your research and
due diligence on these potential partner investors as well. What kind
of companies have they supported in the past? How did they support
them and how could they support you? What kind of expectations
would they have—from the more strategic considerations, such as be-
ing on your board, potential veto rights, and so on, to the more tacti-
cal aspects, such as what kind of reporting are they expecting? Diff er
ent investors have different collaboration models. Some w ill want to
be involved in the day-to-day operations while others prefer to be more
hands-off and wait for a return on investment. Discussions with them
should focus on how they can assist you in validating, executing, and

accelerating your growth strategy, and also on what type of decision
power they expect during the course of operations, rather than how
deep their pockets are. In short, be clear about your needs and what
type of partnership you want with your investors.
What to Know Before Talking to an Investor

Have a robust roadmap. At this stage of your venture development, your
plan should include, for example, your paths to generate regulatory-
grade clinical evidence. Kaia Health, a DTx that focuses on musculo-
skeletal disorders, is a good example of generating early evidence as
the funding came in. The company was launched in 2016 with US$4
million in seed funding. In 2017 and 2018, they published two obser-
vational studies.8 A year l ater, they successfully raised US$18 million
in series A and published outcomes from their first randomized con-
trolled trial (RCT) of 101 patients.9 In 2020, they raised another US$26
million and published their second RCT results with 1,245 patients.10
As in the case of Kaia, generating robust clinical evidence is critical to
achieving a scalable business model and the success of you and your
investor. For an example of how a digital health solution can be
grounded in evidence-based theories and techniques, see appendix B.
Evidence-based DTx platforms tend to be more effective than t hose
that are not, and in short, they can make the difference between a me-
ticulously designed DTx “star” and a hastily put together DTx “flop.”
Stand out from the crowd. As we defined in the first chapter, a DTx is
not a simple mobile app or a digital version of established therapy pro-
grams. T hose could be a starting point, but not the end point. The
importance of demonstrating efficacy through human trials (i.e., RCTs)
goes hand-in-hand with the importance of developing proprietary
solutions. Your startup’s intellectual property (IP) portfolio plays a cru-
cial role in protecting its competitive advantage, in differentiating
from existing or alternative solutions, and in supporting future invest-
ments. As Brian Harris, cofounder and CEO of MedRhythms men-
tioned, “Building a strong IP portfolio is an integral part of our strategy

to build a large, defensible and successful company that makes a big
impact in the world.”11
If DTx can be protected in a manner similar to that of traditional
therapeutics, t hese competitive edges should not be l imited to patent-
able opportunities. A state-of-the-art interface for a unique, engag-
ing, and sustainable user experience, dedicated solutions to caregiv-
ers and physicians, or a unique business model based on a portfolio
approach of complementary DTx solutions can also create unique com-
petitive advantages for DTx organizations.
In short, as you define your growth plan you w ill need to have an
investment strategy to raise capital. Doing your research to determine
what kind of investor your company needs and how they can contrib-
ute to the growth of your venture is critical. The founder-investor re-
lationship is not solely based on money. For investors, the return on
investment is considered an “output” as a result of the relationship (or
investment) with you. They would most likely want to have a say in
the “inputs” and track nonfinancial metrics related to the clinical val-
idation, clinical buy-in, customer acquisition, engagement, and their
experience, as well as the DTx’s overall traction in the market and rev-
enue trajectory. Startup fundraising requires building relationships
with investors, so be prepared to put in time and resources to achieve
this critically important step.
Notes
1. Total addressable market (also called total available market) refers to the
total revenue opportunity available for a product or serv ice. Serv iceable available
market is the total market demand for a product or serv ice, the part of the total
addressable market that can actually be reached. Serv iceable obtainable market,
also referred to as the share of market, is the estimated portion of revenue within
a specific product segment that you are able capture.
2. McGrath and MacMillan, “Discovery-Driven Planning.”
3. Livongo Health, “Livongo Reports Third Quarter 2020 Financial Results.”
4. Teladoc Health, “Teladoc Health Reports Third-Quarter 2020 Results.”
5. “Medtronic Profit Margin 2006–2021,” Macrotrends.
6. McGrath and MacMillan, “Discovery-Driven Planning.”
7. Dilution of control refers to the decrease in existing shareholder-ownership
percentages of a company, also known as equity dilution, which often translates to

a potential lack of control as the founder, entrepreneur, or CEO over your com
pany’s f uture. It’s a concern for entrepreneurs but also a potential red flag for early
investors.
8. Huber et al., “Treatment of Low Back Pain with a Digital Multidisciplinary
Pain Treatment App”; Clement et al., “Implementing Systematically Collected User
Feedback to Increase User Retention in a Mobile App for Self-Management of Low
Back Pain.”
9. Toelle et al., “App-Based Multidisciplinary Back Pain Treatment versus
Combined Physiotherapy Plus Online Education.”
10. Priebe et al., “Digital Treatment of Back Pain versus Standard of Care.”
11. MedRhythms, “MedRhythms Announces an Expansion of Intellectual
Property Portfolio, Securing Patent on Audio Engine for Digital Therapeutics
Platform.”

• Chapter 5
Test, Learn, and Pivot
At this stage, you may have more questions than answers, more sup-
positions than confirmations, about potential therapeutic areas of fo-
cus, the profile of your would-be customers and their motivations,
the price to put on your DTx, and the exact structure of your DTx busi-
ness model. For this reason, it is important as a DTx innovator and
entrepreneur to embrace a culture of “test and learn” from day one.
Commit to refining your envisioned solution and its potential appli-
cation as you gain new insights and user1 feedback.
While you may already have a plan to test, validate, and optimize
the assumptions you have made during the discovery stage, t here are
more steps needed to achieve success in this space. Leading a DTx ven-
ture includes being prepared to face the unknowns (i.e., address gaps or
blind spots in your assumptions that you may not have expected) and
to pivot strategically when required. As business author Leon C. Meg-
ginson says, “It is not the most intellectual of the species that survives;
it is not the strongest that survives, but the species that survives is the
one that is able best to adapt and adjust to the changing environment
in which it finds itself.”2 In other words, the species that survive are
those that are most flexible and responsive to change. If we regard DTx
101

organizations as a “species” of health care organizations, figuratively
speaking, we could say that t hose who adopt a test-and-learn culture
stand the greatest chance of developing g reat products. This culture en-
tails having processes in place that empower teams to continuously and
dynamically optimize their products based on new learnings.
Take for example, Google.com, which exists in many people’s minds
as a source for all answers. The truth is, even within the company it-
self, employees are always searching, testing, and learning how to de-
liver better search outcomes. In one instance, Google was notably un-
able to decide between which shades of blue to use for advertising links
in Gmail and Google Search. This led to a test called, with tongue in
cheek, “Fifty Shades of Blue.” The technology g iant showed 41 differ
ent shades of blue to approximately 2% of its users and found that a
slightly purple shade of blue maximized the likelihood of users click-
ing on the links. Unexpectedly, the slight change in color generated an
extra US$200 million a year in advertisement revenue for the company.
How’s that for continuous learning?
Humana Inc., an American health care insurance provider, is an-
other example demonstrating the importance of testing, learning,
and pivoting. Humana wanted to optimize the first t hing visitors see
when visiting their homepage—t he main banner. Such banners are
important as they create a focus point that captures the interest of vis-
itors with an objective to encourage them to stay on the website for
further browsing, or to click on it to learn about the displayed prod-
uct and ultimately generate purchase o rders. In business jargon, we call
it “lead generation.” Humana ran two tests, comparing a “control” ban-
ner and an optimized “treatment” banner, with different messages,
pictures, and so on, to see which banner had a greater ability to gener-
ate click-throughs. The treatment banners resulted in 433% more and
192% more clicks for the first and second tests, respectively, reflect-
ing a better ability than the control banners to catch more of the visi-
tors’ attention and curiosity in Humana’s solutions.
The type of testing both Google and Humana undertook is known
as A/B testing—a process of controlled experimentation whereby dif

ferent variations of a product or serv ice are served randomly to users
to determine the highest-performing variation.3 If it worked for Goo-
gle and Humana, it can also work for DTx organizations.
What Should We Test?

If your background in product development is in a field unrelated to
health care or DTx, you can still apply your know-how to this new in-
novation terrain. In this case, you should strive to avoid the so-called
curse of knowledge, a cognitive bias when we fail to properly under-
stand the perspective of t hose who do not have as much information
as we do and may require a learning process in order to sufficiently
contribute toward innovation and the productization processes. Deep-
rooted assumptions based on past experience working in another sec-
tor or with a narrow group of constituents do not necessarily translate
into an intuitive understanding of the unique job to be done or how to
address the jobs of the various stakeholders along the DTx value chain.
If you have no background or prior experience in product develop-
ment, and this DTx project is your first entrepreneurship endeavor, you
would want to accelerate your learning curve in order to get up to speed
faster and realize the full potential of your DTx solution.
Regardless of where you stand between t hese extremes on the in-
novation spectrum—whether you are a veteran or a newcomer—
derisking your strategy and business model by testing key assump-
ill be essential.4 Examples of typical assumptions you may
tions early w
wish to test include:
• Would physicians be willing to prescribe my DTx?

• Would patients be willing to pay x amount of money for a
digitally delivered therapeutic?
• How comfortable would users be with a particular feature?
• What role would caregivers play in the care continuum?
• Would payers be willing to implement new risk-sharing
models?
Test, Learn, and Pivot 103

To test your assumptions in an organ ized way, we suggest sort-
ing them across three categories— test now, test next, and test
later—depending on how critical you believe the findings to be for your
strategy and business model.
To illustrate the test-and-learn approach to developing a DTx solu-
tion, let us consider the experience of AEvice Health, a Singapore-based
medtech company addressing chronic respiratory diseases. AEvice’s of-
fering is a smart, wearable patch attached to a child’s chest that acts as a
digital stethoscope to track the child’s respiratory and heart rates during
sleep, which is particularly relevant for managing asthma. Within three
months of starting the company, and with seed investment under
US$5,000, the team launched multiple A/B testing campaigns on social
media to help it figure out and fine-tune the following domains:
• Strategic focus areas: potential “verticals,” or use cases. T

hese
include the applications of the technology to address different
conditions, including asthma.
• Value proposition: optimal strategies to articulate the product’s
potential benefits to different profiles of intended users. For
example, the team explored different taglines, such as “able to
care,” “recording your symptoms,” and “peace of mind.”
• Pricing: finding and testing the sweet spot between a would-be
customer’s willingness to pay for the solution and the investment
the company would need to conduct online user research and
marketing in order to identify and convince potential buyers.
• Call to action: using different marketing campaigns and messag-
ing to encourage website visitors to take the desired action,
such as “subscribe today” or “download now.” In addition,
AEvice tested their website visitors’ willingness to leave their
email address to receive future updates. This action was an
important part of the test as one may like an advertisement but
may not necessarily be interested in buying the product. A
customer leaving their email address is therefore one more data
point t oward gauging user interest.

Addressing Unmet Needs in Asthma
Asthma is a chronic respiratory disease that affected an estimated 262 million
people in 2019 and is the most common chronic disease among children.5 Many
needs of asthma patients remain unaddressed, and there is no gold-standard
diagnostic test and no cure for asthma. The current treatments focus on symptom
management.
Areas of focus for transforming the treatment of asthma are improving patient
education, employing new diagnostic approaches, and implementing personalized
case management.6 Early DTx as “digital companions” propose to help track episodes,
understand the triggers causing the symptoms, and keep patients engaged. Such
solutions—once clinically validated, implemented, and adopted—have a potential to
end the underdiagnosis and undertreatment of asthma, particularly prevalent in
low-and middle-income countries.
The AEvice team obtained thousands of data points from the po-
tential users through the online A/B testing. It also interviewed some
of the individuals who had provided their email addresses. As Adrian
Ang, the cofounder of AEvice said, “None of us are marketing trained.
It was our first step toward d oing market research. T hese experi-
ments helped us refine our targeted customer profiles and their mo-
7
tivations, and also confirmed some of our business model’s assump-

tions.” In short, the research helped the company to understand the
needs and expectations of its potential customers to derisk the jour-
ney ahead.
Understanding how much potential customers are willing to spend
on your solution (a concept known as “willingness to pay”), indepen
dently of your f uture go-to-market approaches and commercialization
strategies (see chapters 11 and 12), merits evaluation. To address this
point, AEvice turned to social media engagement, which then strength-
ened its business case when engaging with potential strategic and
channel partners later in its product journey.

Case Study: A Personal Experience Testing Digital Health Potential for Early
Dementia Screening in Japan
A few years ago, while working with a leading insurance company client in Japan,
a member of our WisDM team was tasked with exploring the potential launch of a
digital health solution for early dementia screening. A traditional route would have
been to conduct lengthy market research that would have cost the client anywhere
between US$50K and US$100K. Instead, with the insurer’s health innovation team,
a test-and-learn approach was used to test the assumptions and local utility of the
concept, which would be much less expensive while yielding more robust insights
compared to the traditional route.
The process started off by examining the typical experience of patients living
with dementia, as well as the experiences of other such relevant stakeholders as
physicians and insurance agents. In interviews with patients, there was a particular
interest in pinpointing the specific moment when they first recognized the
symptoms of dementia and when they got diagnosed, as well as understanding the
emotions associated with those moments. Commonalities across the interviewed
patient journeys to identify recurring pain points were also explored. Armed with
those insights, the team generated over 30 assumptions about patient desire to
understand and assess their dementia risk level, as well as their willingness to take
action to prevent or delay the onset of dementia. Ideating and testing these
assumptions then formed the basis of a potential scalable digital health solution
that spots early signs of dementia.
Test 1: Gauging Interest

The first test launched a Facebook campaign that showed randomly selected users
an invitation to participate in a dementia-related quiz. Similar to Humana’s approach
to optimize their website banner, different messages and images for this campaign
were used to continuously collect such data as click-through rates (CTRs), cost-per-
click, quiz completion, and cost-per-completion.
Within five days on a modest budget, the team collected more than 300 high
dimensional data points. Additional analysis led the team to conclude that Face-
book users in Japan had a strong interest in knowing more about dementia, as
suggested by a fivefold-higher CTR for the messages from the study as compared
to the industry benchmark. The team also concluded that positively framed

messages had higher CTR than negatively framed ones and that potentially
threatening terms such as worried, concerned, and confused should be avoided in
future marketing materials. Additionally, rather than the term dementia, brain health
emerged as the preferred concept as it was easier to spark a discussion around
this topic, particularly with healthy or nondiagnosed individuals.
Through hundreds of quiz responses, the team concluded that in general there
was a lack of understanding and knowledge about dementia. Respondents in Japan
also showed a strong interest in assessing their cognitive risk as early as possible
and were more receptive of words such as preventive rather than reactive. Respon-
dents were also seeking guidance on cognitive risk assessment, such as how to
detect potential early signs of dementia and when it was advisable to see a doctor,
rather than waiting for a dementia diagnosis when preemptive action was possible.
Lastly, the team collected valuable insights on the demographic profiles and
personas of potential customers who seemed most receptive to messaging around
dementia-related solutions. This helped the team obtain a deeper understanding of
the optimal population segment to which a dementia DTx could be targeted. Thus,
test 1 revealed the most relevant assumptions about would-be customers’
concerns around dementia, while also confirming that there was a need in the
market for a dementia diagnostic solution.
Test 2: Monetization Paths

In a second test, the team wanted to explore ways in which an insurance company
could monetize this potential new DTx offering. At the time, most DTx solutions that
fell under the category of value-added services were complementary to core
insurance products, and the team’s goal was to evaluate whether the added value of
a DTx solution would make prospective customers more likely to buy the core product.
The team tested this concept by promoting a DTx solution on social media to see
whether it could generate leads on who would buy an insurance product if it came
paired with a DTx. The team created what is known in product management as a
“painted door test”—an approach that lets developers gauge interest in a product or
service before it’s built. They put together a landing page on the topic of brain
health to validate whether a user would be interested enough in learning more
about the DTx to submit their contact information. If they did, this opened the door
to building further relationships with those users as prospective customers.

Within a few days of the painted door test, the team realized that the CTR on
the landing page was over 10 times higher than the industry standard. Inversely,
the investment required to collect contact information per prospective customer
was 20–60 times lower than the insurance industry average at the time in Japan,
confirming that social media and the deployed mechanisms w
ere a promising
avenue for lead acquisition.
The two experiments pertaining to introducing a dementia screening digital health

solution on the Japanese market demonstrated not only how creative and agile DTx
organizations can be in terms of generating and testing their assumptions, but also
how social media, where appropriate, can be leveraged to do so.
Beyond A/B Testing

Attrition rate (the number of people who initiate an intervention and
later abandon it) is a critical variable for an effective intervention that
requires a sustained user engagement. According to a systematic review
of 22 randomized clinical t rials, assessing the impact of digital health
interventions on health-related outcomes in the workplace, the attri-
tion rates were as high as 60%, with a median rate of 21%, suggesting
that sustaining engagement is a challenge for digital health solutions.8
This realization should lead DTx innovators to ask themselves five key
questions:
• About utility: Does the therapeutic address the unmet needs

and jobs to be done of physicians and patients?
• About usability: Are customers able to use the solution easily?
• About sustainability: Will customers adhere to use by returning
to the solution?
• About desirability: Do customers like the way the solution looks
and feels?
• About branding: What are the customers’ overall feelings about
the solution and brand?

hese questions can be tested in different ways. Below we list four
T
types of strategic experiments you can consider, which are from The
Innovator’s Guide to Growth9 and were utilized in collaboration with In-
nosight, a management consulting company:
• Concept tests are used to pitch solutions to different groups of

stakeholders, using verbal descriptions as well as audiovisual
materials, such as sketches, models, and video clips of the
hypothesized user experience. The audiovisual elements help
stakeholders visualize the potential concept and bring the idea
to life, enabling them to give more accurate feedback on which
solutions are likely to work and how they can be improved. A
concept test could also show mock-ups of a DTx interface to
gather qualitative insights from potential users.
• Functioning prototypes are tangible and experiential models of
the envisioned solution. By observing how stakeholders inter-
act with them, and assessing features that may drive sustained
interaction, developers can gain a better understanding of how
well the proposed solution and key features deliver against the
jobs to be done. Feedback and learning can be captured through
iterative stakeholder interactions as the prototypes get increas-
ingly refined.
• In transaction tests, the offering is presented to customers for
consideration and purchase in a setting that mimics the
natural purchase environment. T here is no better way to
substantiate the viability of a new business model than recreat-
ing real-world transactions to determine if customers would
pay for the solution.
• Operational pilots are used to provide real in-market learning
within a scaled-down environment. It allows the team to vali-
date some of the most important assumptions around market-
ing, distribution, repeated purchases, and resource require-
ments, which are hard to test u ntil the solution is made available
to a big enough market for a long enough period of time.

Go Further with Multivariate and Factorial Testing
A range of test designs can shed important insight on the mechanisms driving user
interactions with DTx. Compared to A/B tests, multivariate tests compare a higher
number of variables, revealing insights that can only be gleaned from understand-
ing how these variables interact with one another. In addition, factorial experiments
can infer interaction spaces between the tested factors by assessing a small
number of carefully selected combinations. For example, a factorial design experi-
ment was used to identify effective intervention components for a smoking
cessation program. The study tested six intervention components: nicotine patch
vs. no intervention (none); nicotine gum vs. none; counseling vs. none; intensive
cessation in-person counseling vs. minimal; intensive cessation phone counseling
vs. minimal; and 16 vs. 8 weeks of combination nicotine replacement therapy.10
Through this test, performed with carefully selected combinations of factors, the
researchers concluded that preparation counseling and the combination of
intensive cessation in-person counseling with preparation nicotine gum or patch
were promising intervention components for smoking and should be evaluated as
an integrated treatment package.
As the examples from Google, AEvice, and the dementia case study
in Japan show, for any of the tests mentioned here, and for product
development overall, a recommended practice is to develop and test
the assumptions with the would-be end users. Designing a product
that is wanted by its user base can then be built using the test-and-
learn approach and the following best practices:11
• Design the product using a human-centered approach, accounting for

the user’s core needs, abilities, and environments of use, and device
interface.
• Design the product to optimize adoption, engagement, and adher-
ence within the target population.
• Integrate product accessibility and universally designed features
where appropriate.

• Conduct h
uman factors testing throughout the product development
and usability processes.
From ideation to commercialization and post-marketing software

upgrades, the most impactful and enduring DTx solutions will be those
that undergo constant improvement based on input from the larger
deployment ecosystem. This ecosystem, as well as its criteria for ad-
herence, can be dynamic, changing over time both across users and
even for the same user. Hence, a test-and-learn culture throughout the
organization should be part of any ambitious DTx company’s DNA, as
proven by the most advanced technology firms. For this feedback loop
to work, DTx innovators must therefore build mechanisms that enable
frequent testing and evaluation into the development cycle of their
product and maintain t hese processes from prototyping all the way
through to post-market release phases.
Conclusion
When piloting DTx solutions in day-to-day care, it is critical to con-
tinuously test and validate the solution itself and its business model
with evolving use contexts. The tools, behavioral models, and tech-
niques presented in this chapter should be a starting point to help
your organization overcome some of the most common challenges
faced by DTx organizations at this stage and get you on track to move
from pilot to production.
Notes
1. We prefer the term user, or end user, over patient to refer to the person who is
ultimately going to use the DTx as it could be patients themselves or caregivers,
physicians, or any health care professionals.
2. Megginson, “Lessons from Europe for American Business.”
3. Young, “Improving Library User Experience with A/B Testing.”
4. There are risks inherent in launching any venture. They can be risks related
to the product, your business model, the team you are going to hire, investors you
are to let join the board, and so on. Being involved in a DTx venture w ill come with
specific risks and uncertainties related to the nature of the technology evolving in

a complex environment. New pathways need to be created (e.g., regulatory and
reimbursement approvals) and many questions have not being completely
addressed yet (e.g., physicians or patients buy-in for such new solutions). There-
fore, an important part in being successful in developing, validating, and imple-
menting a DTx is about managing t hese risks, trying to define, analyze, and
anticipate them. In short, it’s about derisking your venture’s strategy and overall
business model. As we use this term, derisking, throughout the book, we w ill
suggest how others have done it and suggest recommendations.
5. World Health Organization, “Asthma: Key Facts.”
6. World Health Organization, “Asthma: Key Facts.”
7. This is independently if you are going the business-to-business (BTB) route,
where a health care practitioner w ill prescribe the DTx solution to a patient or a
business organization w ill make your solution available to the end user (e.g.,
employers for their employees), or the business-to-consumers (BTC) route, where
you are serving directly the end user of the solution, through an online e-
commerce platform, for example.
8. Howarth et al., “The Impact of Digital Health Interventions on Health-
Related Outcomes in the Workplace.”
9. Anthony et al., The Innovator’s Guide to Growth.
10. Piper et al., “Identifying Effective Intervention Components for Smoking
Cessation.”
11. Digital Therapeutics Alliance, DTx Product Best Practices.

• PART III
FROM PIL OT TO PRODUCTION
Thinking beyond the Algorithm

Our ethos at WisDM states, “Technology alone cannot change health
care.” It is a particularly important consideration when the technol-
ogy is transferred from proof of concept into standard of care. As
Dr. Carolyn Lam pointed out in the introduction, “It’s not b ecause you
have an algorithm that you have a commercial-ready product. We
should not underestimate the amount of work that it takes to inte-
grate with existing electronic medical records, fix data silos, design
the right interface, protect patient data, and get all the regulatory
approvals.” During this part of the DTx journey, it is important for
DTx entrepreneurs to have not only the “right” algorithm but also the
“right” product that can be integrated successfully into everyday
clinical practice.
In this section of the book, we walk you through the steps required
to overcome common DTx implementation challenges, starting with
the importance of addressing aspects of data sourcing and manage-
ment (chapter 6). We also touch upon the critical importance of
generating evidence and exploring novel approaches to clinical valida-
tion (chapter 7).
Variability in Care Delivery

Pilot studies aim to preliminarily examine specific variables of
interest pertaining to a candidate intervention. T hese preliminary
studies can determine if larger, sufficiently powered studies are
warranted, or to potentially refine trial designs should the study be
113

expanded. Pilot studies are usually conducted u nder controlled
conditions, which can help to isolate and generate conclusions about
these variables. However, results from pilot studies may not fully
represent all the nuances and circumstances of delivering care to
patients in a routine, real-life setting—often termed real-world data.
It is important for DTx entrepreneurs to understand this limitation
and why interventions with promising outcomes from pilots do not
always translate into routine clinical practice.
“Many promising technological innovations in health and social
care are characterized by non-adoption or abandonment by individu-
als, or by failed attempts to scale up locally, spread distantly, or
sustain the innovation long term at the organization or system
level,” says Trish Greenhalgh, a British primary care practitioner,
academic, and one of the world’s foremost experts in research
translation.1 This challenge expands much deeper into the complexi-
ties of clinical practice beyond an introduction of a new technology.
Even with the highest level of evidence backing the proposed solu-
tion, cementing change in clinical practice is not easy. Let’s consider
two distinct studies looking at adherence to national guidelines—
that is, best clinical practice recommendations based on evidence
and agreed on at a national level to lead to the most optimal results.2
In a study performed in the United States, Dr. Elizabeth A.
McGlynn and her team of researchers telephoned a random sample
of over 20,000 adults over a two-year period and asked them about
their health care experiences. The researchers also reviewed the
patients’ medical records and evaluated the quality of care the
patients had received using 439 indicators for 30 acute and chronic
conditions (such as diabetes, hypertension, and cancer) as well as
preventive interventions (such as mammography screening for
breast cancer, screening for cervical cancer, or having smoking
status documented). The study found that only 55% of patients
received the type and level of medical care recommended by national
guidelines3 for their conditions, with substantial variability between
conditions, ranging from 78.7% of recommended care for senile
114 From Pil ot to Production

cataract to 10.5% for alcohol dependence.4 In short, the study re-
vealed a substantial gap in the care as recommended to be provided
and as received by the individuals, suggesting suboptimal adherence
to standard health care guidelines by the medical professionals.
Suboptimal adherence to standard health care treatment guide-
lines is not specific to the United States. Another recent study
performed in Australia examined whether health care for c hildren
aged 15 years and younger was consistent with established quality
standards. The study assessed the clinical records for 17 high-
incidence child health conditions (such as asthma and type 1 diabe-
tes) and estimated the overall adherence to be 59.8%, with substan-
tial variation across conditions. Taking asthma as an example,
guidelines for the management of asthma recommend that each
child receive a written action plan. This recommendation was
followed for approximately 47% of the c hildren assessed by this
study who were prescribed an asthma inhaler.5
Awareness of a variability in provider adherence to national
guidelines and evidence-based practice overall is a starting point to
understanding f actors that influence the implementation of new
technologies and why beyond pilots, full-scale adoption of technolo-
gies, including DTx remains a challenge.
Implementation Science
As we stated earlier, our innovation ethos is that technology alone
cannot transform health care. Beyond the ideation and validation of
promising technologies, t here remains a critical need to develop a
road map toward successful deployment of t hese ideas into clinical
practice to catalyze broad adoption across providers through patient
communities. A fter all, highly innovative, evidence-backed technolo-
gies that cause inconveniences to clinical workflows may end up not
being a dopted. This is where domains such as implementation
science (IS), defined as the scientific study of methods to promote
the systematic uptake of research findings and other evidence-based
From Pilot to Production 115

practices into routine practice, become essential. Many successful
scale-up programs are based upon the principles of IS,6 which
provides a useful lens to understand key considerations for introduc-
ing DTx into new care delivery settings that should be top-of-mind
for DTx entrepreneurs and innovators.
Based on our own experience, interviews with digital health and
DTx CEOs, and the published evidence, implementing and sustain-
ing DTx (or any technology) in health care is a complex undertaking.
We therefore look at leveraging the factors that influence the suc-
cessful implementation of new technologies in chapter 8. We w ill
also offer a how-to approach to implementing DTx, following the IS
approach and offer the perspective of health care system managers
on DTx integration in chapter 9. Finally, we discuss how DTx entre-
preneurs might use behavioral methods like “nudging” to gain
clinical buy-in and patient acceptance in chapter 10.
Technology-driven innovation moves at an extraordinarily fast pace,
much faster than a health care system’s readiness to integrate this
technology into existing infrastructures, processes, and cultural
norms. This misalignment applies to DTx too. We believe that during
the transition from a DTx prototype to pilot study, and ultimately to
commercial product innovation and real-world data generation, the
innovators should be guided by a structured approach to identify and
address the substantial challenges of changing clinical practice at scale.
Notes
1. Greenhalgh et al., “Beyond Adoption: A New Framework for Theorizing and
Evaluating Nonadoption, Abandonment, and Challenges to the Scale-Up, Spread,
and Sustainability of Health and Care Technologies.”
2. Frequently used as a measure to determine the quality of care.
3. Guidelines are used for clinical decision-making and reduce practice
variation to ultimately improve patient outcomes.
4. McGlynn et al., “The Quality of Health Care Delivered to Adults in the
United States.”
5. Braithwaite et al., “Quality of Health Care for Children in Australia, 2012–2013.”
6. Greenhalgh and Papoutsi, “Spreading and Scaling up Innovation and
Improvement.”

• Chapter 6
Data Sourcing and Management
Collecting and managing medical data is not a trivial task. It requires

substantial infrastructure, capital, expertise, trust, and time. You will
experience the complexity of working with medical data throughout
your DTx journey. The first touchpoint—data sourcing—is when you
source medical data to test and demonstrate the basis for your idea.
The second touchpoint—data management—is when you have the data
under your purview, either sourced externally or collected by your
platform.
When you approach a dataset owner to negotiate access to the data,
be prepared to discuss the extent of that access and who w ill own the
data processing products (that is, your analytical tool and the data
analysis results). Additionally, once you start analyzing the data, you
are responsible for the data stewardship—ensuring the privacy of the
individuals who contributed to the dataset and the security of their
information. Once your platform starts collecting and generating the
data itself, you need to ensure that appropriate measures are in place
to do the same for the newly collected data.
Questions spanning not only technology but also ethics and legis-
lation w
ill certainly arise throughout your quest to secure access to the
117

type of medical data on which your DTx w ill run and to identify the
right framework and processing rules for managing t hose type of
sourced and newly collected data. In this chapter, we provide you with
a detailed overview of the most common medical data sources and out-
line key considerations around data ownership and stewardship.
DTx-Relevant Data and Where to Find Them

One of the most common types of data used for training and testing
medical software is deidentified retrospective medical datasets,
which contain patient medical history (diagnoses, visit notes, and
treatment courses) with such patient identifiers as names and social
security numbers removed. Aspiring DTx startups have several options
to obtain access to t hese datasets:
1. Access theme-specific, publicly available data repositories (also

known as data commons), which may or may not include accom-
panying analytical tools and are open for anyone to use in the
spirit of open science. T hose datasets are often the output of
research projects that have been funded with the requirement
to publish their data processing products and results.
2. Participate in private (e.g., corporate-or insurer-sponsored) and
public (e.g., hospital-or university-organized) innovation events
and incubator programs specifically aimed at startups that enable
them to access data governed by t hese institutions. Of note:
participating in such events may require physical attendance at
the institutions’ premises (e.g., institutional incubator programs).
3. Join data consortia that allow for data usage without interfer-
ing with ownership or governance (e.g., multiparty computation
networks). This can usually be done remotely but w ill likely be
subject to consortia-specific guidelines and contracting.
4. Buy access to medical data (e.g., for-profit datasets). In t hese
cases, data sourcing and ownership should be carefully vetted
to ensure their ethical sourcing, quality, and reliability.

As you can see, there are multiple sources where one can find medical
data. However, while finding a data source may not seem too challeng-
ing, finding the right data source is a different story. What is the dif-
ference between the two and what it does it mean for you?
How to Identify a High-Quality Data Source

The right data source should provide data of high quality and relevance
to the patient population your solution is designed for. This is because
your data processing products (DPP)—which comprise the tool, soft-
ware, and technology developed and refined with the data found in
your chosen data source, and the end results of the data analysis per-
formed with that tool—w ill ultimately serve as the evidence backing
your idea, and they can be only as good as the quality of the data at
their foundation. Therefore, if the data are of poor quality, so will your
DPP. These considerations are echoed in our conversation with Dr. Kee
Yuan Ngiam, Group Chief Technology Officer of the National Univer-
sity Health System in Singapore1—a key type of stakeholder with
whom a DTx organization may engage toward the deployment of its
platform at a large health care system: “Our collaboration with poten-
tial partners such as DTx, assuming they have demonstrated that they
are safe and effective to be used by our clinicians and patients, starts
with a discussion about data. It’s important for us to understand the
data sources—type and quality of the data sources that have been used
to feed and train the model, the context these data have been derived
from, and would it be applicable to our local context here, at the hos-
pitals where the technology might be deployed and used.” Read more
from Dr. Ngiam on best data practices for implementation in chapter 9.
Similarly, to ensure the integrity of any data-driven initiative, it is
crucial that the data in question are legally sourced. Data provenance—
the records on how and where the data w ere collected and what has
happened to them since—inform to a g reat degree the quality, authen-
ticity, and credibility of the data. T hese factors substantially influ-
ence everything from the adequate training and external validation
Data Sourcing and Management 119

of AI algorithms to how patients can be stratified for treatment, among
other critically important parameters. Importantly, datasets can also
serve as a source of bias, a major challenge that can adversely impact
the health outcomes of patients who w ere underrepresented in the
datasets, for example patients with rare medical histories, or patients
from low-and middle-income countries. Bias—a systematic error in
preference of an idea vs. its alternative—is present in both human
and algorithmic decision-making. Bias in medicine has a far-reaching
consequence on the individual and population health outcomes.
The first step in overcoming biased decision-making is the under-
standing of where it is coming from. In case of the algorithmic decision-
making, the first place to check for bias is the underlying data used by
the algorithm. A reliable and trusted data source does not guarantee
the data’s relevance to the population you are trying to help. For ex-
ample, a dataset of otherw ise good technical quality (i.e., consisting
of thousands of patients, well organized, traceable, and with an accept-
able level of missing data fields) but in which the average member is a
55-year-old white male with a body mass index (BMI) of 35 can lead
to a biased interpretation of an average diabetes patient’s clinical pro-
file and influence the training data used in the development of the al-
gorithm of a hypothetical DTx solution for diabetes. As a result, an
algorithm built on such data may trigger automated insulin release at
a dosage calibrated based on recommendations for a person with this
particular physiological profile, and that could have serious implica-
tions if the same dosing were applied for a 21-year-old Asian female
with a BMI of 18. To add even more complexity to the dosing problem,
optimized insulin dosing parameters for certain patients may be highly
individualized and not be directly correlated to traditional consider-
ations such as body weight and/or BMI. In this scenario, an insulin
overdose for the younger female could lead to hypoglycemia—a glu-
cose level that is too low—w ith potentially serious outcomes, includ-
ing loss of consciousness in extreme cases.
If an analytical tool is developed based on data from a clinical trial
that used excessively rigorous inclusion and exclusion criteria, whereby

only “ideal” patients were recruited (that is, people with no comorbid-
ities), it may have limited application to patients in real-world situa-
tions, where disease symptoms may be less clear-cut and one’s clini-
cal biomarkers may reflect the impact of several intertwined health
conditions. To avoid such missteps, the datasets used for developing,
training, and validating algorithms, such as t hose that underpin DTx
technologies, should represent patient populations that are as similar
as possible to the ones the solutions intend to serve. And even then,
one should keep in mind that data are not the only source of bias. The
analytical tools, if incorrectly designed, can skew the decision based
on the otherw ise representative data. This can occur, for example, by
applying a higher importance to predefined, biased factors. Being
aware of the existence and risk of biases, in both datasets and the al-
gorithms themselves, allows developers to incorporate design features
that can limit them (e.g., fairness-aware machine learning) and for the
overseeing body to put in place an appropriate framework to monitor
and minimize any potential bias.
Who Owns the Data Processing Product?

The crowning achievement of your work is the product of your data
processing. Collecting data can take years and requires a considerable
investment in infrastructure and a skilled workforce, hence the op-
tions we presented to you on where you can find ready-to-use data-
sets. Once you have identified and gained access to the optimal data,
you can use it to develop, validate, and refine your tool. At one point
you w ill start collecting the data yourself and may even start forget-
ting about the challenges of your early search for the optimal dataset.
Before that happens, though, t here is an important question to be an-
swered: who owns the product of your work given that all of the data
upon which it is based has been laboriously collected by someone e lse
from yet another party (the patient)?
The complexity of this question increases as the number of parties
involved that contribute either data or design and technological

know-how, or both, grows. “It is never too early to have a conversation
with your business partners and collaborators about the ownership
of the data and the data processing products,” says our contributor
Dr. Georgios Kaissis, senior research scientist at the Institute of Arti-
ficial Intelligence in Medicine and Healthcare and leader of the health
care and research units at OpenMined, an open-source community de-
veloping privacy-preserving machine learning tools and training.2
This conversation may not be easy, and it is helpful to refer to legal
frameworks if they exist. Most common, the intellectual property (IP)
rights—including the ownership of your DPP—may have to be speci-
fied on a case-by-case basis.
Nicholas Brocklebank, director of in-house legal digitization and
transformation at the Global Legal Solutions Group, a provider of legal
operations solutions and dedicated services for startups and the man-
agement of IP, points out that datasets are generally considered to be
the intellectual property of the party that created them. So, to gain
access to such datasets, you w ill require a license from that owner, the
terms of which w ill be whatever you can commercially agree on with
that owner.
What does this mean in practice? While some medical and re-
search institutions charge a nominal fee, or no fee at all, for access
to their data as part of their mission to support medical technology
development, the parties owning the data are f ree to set any price
they consider appropriate. It is not uncommon to see some organ
izations require e ither costly payments for access or ownership rights
in any DPP that is developed with use of their data. Brocklebank
tells us,
The key to avoiding unpleasant surprises is to read the terms of any license
agreement in detail before proceeding to access the dataset (easier said than
done, as so many of us are used to clicking through stuff online, but this is
really important). The upside of carefully reading license agreements is that if
you consider the cost or terms to be excessive, but you still find the data
valuable for your project, you can try to negotiate by reaching out to the data

owner—or simply look elsewhere for a similar dataset with more favorable
access terms.3
Lastly, if the resulting technology is to be commercialized and mone-

tized once it has been developed based on the licensed data, the parties
will have to agree on profit sharing. It can get complex quite quickly if
the project involves multiple parties providing data and/or technical de-
velopments. “One possible approach is to implement an infrastructure
that tracks what data were used and what code was developed by each
team so as to evaluate the contributions of each party, considering both
quantity and quality,” shares Dr. Kaissis. It may sound like a hassle, but
with the global DTx market projected to reach US$11.82 billion by 2027,
it is imperative to treat the ethical, legal, and technical foundations of
this thriving ecosystem with the highest attention to detail.4
Data Privacy and Safety

With an increase in the digital literacy of current and future technol-
ogy consumers, the user of digital technologies has a growing aware-
ness of the value of their data and the threats related to data breaches.
The demand for the increased control over one’s data access has been
heard by the regulators and by the tech companies. In 2018 the Euro
pean Union established its comprehensive privacy protection law,
known as the General Data Protection Regulation, signaling a readi-
ness on the part of national and international regulators to step in
and demand an increased control of data privacy and anonymity.
Shortly thereafter, in 2019, the New York Times released a collection of
articles on privacy titled The Privacy Project that featured an opinion
piece by Google CEO Sundar Pichai (“Privacy Should Not Be a Luxury
Good”) in which Pichai outlined Google’s strategies for data manage-
ment that increased privacy.5 Another technology giant, Apple, used
privacy as a selling point for its iPhone 12 ad campaign highlighting
the iPhone’s App Tracking Transparency feature to manage data access
when using the phone.

The desire for privacy (or the fear of losing it) can have real impli-
cations for the adoption of health care solutions, even if these solu-
tions have a huge potential benefit. A good example of this scenario
has been the reluctance of the general public to install COVID-19
contact tracing apps, as shown by a previous study assessing popu-
lations in Australia, France, and the United States.6 Importantly,
privacy concerns are only increasing and its consideration in the de-
velopment and downstream use in DTx w ill need to be continuously
addressed.
What does this mean for your DTx endeavor? Understanding pri-
vacy expectations and designing your product with built-in privacy
protection mechanisms w ill help you not only avoid security and reg-
ulatory missteps but also minimize the risk of nonadoption and avoid
issues of noncompliance down the line. To do this right, we recommend
engaging a cybersecurity or privacy expert as soon as your technology
passes the initial data collection, algorithm training, and validation
stages. H
ere we highlight several high-level considerations for DTx
platform deployment into the real world.
Privacy Threats
A key first step t oward protecting privacy by design is identifying the
threats. “Even if you have the best intentions in mind, if the data is
attractive, it is under threat,” says Professor Reza Shokri, a researcher
of data privacy and trustworthy machine learning from National
University of Singapore’s School of Computing. The risk of privacy
threats can be divided into direct and indirect information leakages.
Direct information leakage occurs within an organization, for exam-
ple, by breaking contextual integrity (using the data for a different
application than what the data donor consented to). Indirect data
leakage occurs when, based on the DPP or on data introduced into the
system with an adversarial objective, one can reconstruct and deiden-
tify the underlying dataset.

The Threat of Reidentification
here is a common belief that removing key identifiers, such as a per-
T
son’s name or identification number, is sufficient to ensure privacy. Un-
fortunately, this step is far from sufficient. Even more unfortunate is
the fact that it may not guarantee protecting against patient identifi-
cation, especially when those data are of value. However, knowing full
well that more steps beyond removing key identifiers may be needed
to effectively prevent patient reidentification, this may lead to ques-
tions of w hether or not the resulting dataset w ill become so redacted
that it is no longer useful.
One may also wonder how it is possible that data contributed by one
individual cannot “hide” in the tens of thousands of other data points
to naturally protect against reidentification. The explanation is that
while a single attribute, such as gender or location, is not a key identi-
fier and typically cannot be traced to a particular person, multiple at-
tributes across high-dimensional datasets that are associated with the
same person, such as gender, age, and multiple check-in locations, can
be easily triaged by a skilled malevolent agent to reidentify that indi-
vidual. These points illustrate why the issue of data security w
ill always
serve as a major consideration for companies in the digital health space.
Want to Go Further?
Data security and privacy preservation are rapidly and dynamically evolving fields.
As the scholarly publications on these topics are frequently updated, we recom-
mend following the work and educational resources of open-source communities,
such as OpenMined, and leading organizations in the field, such as the Simons
Institute at the University of California, Berkeley. It may also be helpful to explore the
latest updates at key scientific and industry meetings, such as the Healthcare
Information and Management Systems Society Global Health Conference, the
Conference on Neural Information Processing Systems, and the Conference on
Computer and Communications Security, among others.

Tools to Provide Data Safety
Direct information leakages can potentially be avoided by continuous
evaluation and reevaluation of the strategies being employed by an
organization to protect its data infrastructure. This can be further
bolstered by robust auditing processes to assess the data-access work-
streams within an organization, and this oversight applies to both
raw and processed data. To address reidentification and other forms
of indirect information leakage, various approaches and tools are be-
ing developed to ensure that both data processing and results pub-
lishing are done with safety features that protect the privacy of the
individuals who have contributed their data. Some of those tools can
be implemented at the level of your DTx platform, while o thers re-
quire buy-in from the broader community and the stakeholders with
whom you want your technology to interface (e.g., a hospital or a health
care system).
As in many other fields, data are a critical component of the validation

of eventual deployment of medical innovation. For that reason, data
should be viewed as both a vital business asset as well as highly sensi-
tive personal information, provided by patients and users for the ben-
efit of the global community. Responsible data sourcing and manage-
ment entails safety, ethical, and legal compliance, and the principles
of responsible data management should be integrated into emerging
DTx platforms at the very start of their life cycles.
Key Takeaways
1. Datasets can be obtained from a range of sources. Whichever source or
sources you choose to drive the development of your DTx platform should
be carefully vetted since it w
ill form the foundation of your innovation.
2. Data ownership and rights to the data processing product must be consid-
ered as early as possible in the innovation road map.

3. Privacy preservation is a critical consideration from ethical, legal, and
adoption standpoints. Invest your time in implementing data management
framework(s) that are appropriate and adaptive toward evolving needs and
standards in data security and privacy protection.
Notes
1. National University Health System, one of the three public health care
clusters in Singapore, is composed of several hospitals, national specialty centers,
and polyclinics. More from Dr. Ngiam on the stakeholder expectation can be found
in chapter 9.
2. From an interview with the authors, December 10, 2020.
3. From an interview with the authors, January 12, 2020.
4. Precedence Research, “Digital Therapeutics Market Size to Hit US$ 11.82 Bn
by 2027.”
5. Pichai, “Privacy Should Not Be a Luxury Good.”
6. Chan and Saqib, “Privacy Concerns Can Explain Unwillingness to Download
and Use Contact Tracing Apps when COVID-19 Concerns Are High.”

• Chapter 7
Clinical Validation
Unique Challenges and Novel Approaches
This chapter was coauthored with Jennifer Shannon, MD, CEO at

Digital Brain Consulting, clinical advisor to several digital therapeutic
companies, Regional Medical Director at Cognoa, and a member of the
Digital Therapeutics Alliance Advisory Board; as well as Xavier Tadeo,
PhD, clinical trial and regulatory affairs specialist at the N.1 Institute
for Health at the National University of Singapore.
Digital therapeutics, like drugs, need to be established as being safe and

effective. And much like the work that goes into bringing new medi-
cines to market, supporting t hese objectives requires marshaling scien-
tific evidence that attests to the clinical validity of these technologies.
First, the DTx must be validated using randomized controlled trials
(RCT). Other research modalities commonly used for testing DTx are
observational studies (where the researcher observes processes and
outcomes without intervening or assigning treatments, and the results
can be descriptive or analytical), real-world outcomes studies (research
comprising real-world data to assess health outcomes and advise health
care management), real-world evidence studies (analysis of real-world
data obtained from retrospective or prospective observational studies),
128

and economic evaluation studies (analysis of financial impact on the
patient and payer via cost effectiveness analysis). These forms of anal-
yses include considerations of quality-adjusted life year (QALY), or the
cost to prolong life while accounting for the quality of life. Additional
analyses may look at the equal value of life years gained (evLYG),
which primarily assesses the degree to which life is lengthened irre-
spective of the quality of life gained. T hese strategies have served as
pillars for how interventions can be assessed when their deployment
is being considered.1 Importantly, as the application of DTx continues
to expand for such domains as chronic disease prevention and man-
agement, oncology, and other indications that generate substantial
cost burden on health care systems and patients, the aforementioned
measures represent both standards and calls to action to which DTx
platforms have the opportunity to meet.
Studies that involve any identifiable health information or human
biological material or that subject an individual to any intervention
must be carried out under the supervision of an institutional review
board (IRB) or other applicable institutional ethics/medical board and,
in many cases, accompanied by a recommended listing on a publicly
available registry, such as ClinicalTrials.gov or clinicaltrialsregister.eu,
upon study launch. A fter the product is granted market access, evi-
dence must still be collected and analyzed on an ongoing basis to sup-
port the product’s effectiveness and safety assurances in a real-world
setting and to possibly give insight into new, optimized iterations of
the product. In a nutshell, if you are a DTx developer or entrepreneur,
there are substantial bars to meet before a product can be considered
for regulatory clearance, subsequent clinical deployment, and receipt
of reimbursement codes from insurers where applicable.
With the emergence of digital health in the marketplace, there
has been a corresponding increase in industry-academic partner-
ships to drive evidence generation. As with traditional research and de-
velopment (R&D), DTx developers can and should collaborate with
academic research centers and hospitals to facilitate and potentially
accelerate many of the clinical and regulatory requirements for DTx
Clinical Validation 129

commercialization. Some of the key advantages at the foundation of
these partnerships include the ability to integrate the go-to-market
framework of an industry-developed DTx with the rigor associated with
IRB review and approvals, potential for collaborative clinical trial de-
sign and biostatistical method innovation, and ultimately, the codevel-
opment of peer-reviewed publications that serve as a bedrock for due
diligence and independent evaluation of research evidence. Contract
research organizations (CROs) can provide similar services, and part-
nering with them can be particularly useful in setting up multisite trial
recruitment since data monitoring from numerous sites can be insur-
mountably challenging to manage without substantial infrastructure
familiar with the regulatory environment in place. In addition, CROs can
provide instrumental insight in later stages of the clinical validation of
a DTx, when other stakeholders beyond trial participants are engaged.
Besides being established as effective for patients (users), and com-
plying with legal requirements as well as ethical considerations, the
clinical validation of a DTx represents a major inflection point when
it comes to convincing potential investors to back a venture. Without
validation, apps are subject to classification within the wellness cate-
gory, an overcrowded space where standing out is a challenge. How-
ever, validation is expensive. To solve this paradox, entrepreneurs can
potentially turn to grants and angel investors during certain stages of
development prior to exploring the venture capital route.
Regardless of the funding route taken, investors and some grant-
ors w ill require varying degrees of assurances, objectives, or other
types of key performance indicators. Value metrics, such as the ones
provided by the leading professional society for health economics and
outcomes research (the International Society for Pharmacoeconomics
and Outcomes Research, or ISPOR),2 can help decision-makers to un-
derstand hard-to-measure benefits and costs (economical and clinical)
of a therapeutic intervention.
The journey to DTx clinical validation presents some unique chal-
lenges not found in the traditional R&D process. H ere we review eight
of them in detail.

Defining Digital Clinical Endpoints
Clinical endpoints are assessment criteria that measure the safety, ef-
ficacy, and overall effect of an intervention (e.g., drug or therapy) in a
clinical trial. T
hese can refer to survival rates, occurrence of disease,
biomarkers, evolution of symptoms, or drug toxicity, just to name a
few. The five-year survival rate of stage IV cancer patients, decreases
in viral load, and the frequency and severity of adverse events are some
examples of clinical endpoints.
Digital endpoints, also known as digital biomarkers, are a subtype
of clinical endpoints based on physiological and behavioral data tracked
and measured digitally, often by the investigational device itself. De-
vices that collect highly accurate data from which such endpoints can
be derived include accelerometers, activity monitors (also known as fit-
ness trackers), digitally embedded microphones, and blood pressure
sensors, to name a few. Examples of digital endpoints include heart
rate, breathing rate, blood oxygen saturation, changes in speech pat-
terns, or sleep duration and quality.
The unique value of digital endpoints for research is that they may
augment the quality, quantity, and granularity of data collected in a
trial. Because they typically do not require physical measurements by
a nurse or a nurse technician and reduce the need for in-person visits
to clinical study sites, they are likely to make clinical t rials in which
they are incorporated more attractive for potential participants. In ef-
fect, these trial designs and associated technologies can decentralize
clinical trial execution, where participants are monitored remotely
through automated dataflow. This approach has seen a substantial
increase following the emergence of COVID-19. T hese decentralized
trials in turn may drive patient enrollment and adherence, benefit-
ing both patients and clinicians by freeing up time and eliminating
trips and interactions that are not absolutely necessary. Indeed, in
decentralized or hybrid trials, digital endpoints are indispensable,
particularly when trial participants hail from rural areas or other low
population density settings. T hese designs can dramatically advance

observational and longitudinal studies by collecting data over time
with markedly reduced workload and personnel required compared to
traditional approaches. This in turn makes them an ideal instrument
for the collection of real-world data, such as t hose generated during
phase IV (post-marketing) pharmacovigilance studies.
In terms of therapeutic areas, digital endpoints are most useful for
investigating conditions where human supervision can be partially au-
tomated or where clinical judgment and shared decision-making can
be supported by algorithmic analysis. Hypertension, diabetes, and
more “subjective” disciplines, such as psychiatry and neurology (which
are largely based on observational assessments) are a few examples of
areas where the application of digital endpoints can improve the qual-
ity of patient monitoring.
Companies interested in leveraging digital clinical endpoints through
technology face some important entry barriers, however. Firstly, any as-
sociation between a digital biomarker and a specific condition must be
backed by solid clinical evidence. A notable challenge in this instance is
showing that changes in the digital biomarker correlate with disease
state in patient clinical outcomes. In other words, output data from the
DTx using a digital biomarker need to appropriately reflect the clinical
concept of interest. Moreover, the accuracy and reliability of digital bio-
markers needs to be assessed against current standard biomarkers be-
fore regulatory approval; output data from the DTx using a digital bio-
marker needs to be accurate when compared to a gold standard.
Digital endpoints and biomarkers are relatively novel and a compre-
hensive regulatory framework that governs their adequate usage is
yet to be developed. Thus, apart from developing, testing, and honing
the hardware and software of their solution, DTx companies also need
to establish and validate what they consider to be relevant endpoints
for the therapeutic area in which they have chosen to operate.
In 2019 several pharmaceut ical companies that sponsor clinical
trials decided in the spirit of collaboration to share their intellectual
property to create a public digital endpoint database, now available at
the website of the nonprofit Digital Medicine Society (DiMe).3 This

database is applicable not only to DTx, but to all type of interventions
whose effect is to be measured with digital biomarkers. This crowd-
sourced resource allows such key stakeholders as principal investiga-
tors, DTx developers, and collaborating clinicians to search for stan-
dardized, validated clinical endpoints when developing digital health
technologies. To cite one example, UCB Biopharma, a contributor to
the DiMe database, shared the digital endpoints the company used in
a phase IV trial to assess the efficacy of a wearable technology it had
developed for improving patient adherence to the drug Neupro, com-
monly prescribed to mitigate Parkinson’s disease symptoms.4
Meanwhile, for t hose who still prefer to develop their own digital
biomarkers, the Clinical T rials Transformation Initiative (CTTI) has
created recommendations and best practices for identifying, selecting,
and developing novel clinical endpoints.5 Cofounded by Duke Univer-
sity and the US Food and Drug Administration (FDA), CTTI is a group
of more than 500 organizations devoted to driving the adoption of
optimal trial design and execution practices to modernize research
strategies among its community.
Comparing DTx to Standard of Care

Traditionally, randomized clinical t rials follow a design where patients
are allocated into one of two or more groups that can include control
and treatment groups. Patients randomized into the control group re-
ceive either the standard of care or a placebo. As the standard of care
is well defined,6 this allows the experimental treatment to be compared
to the control group in order to assess its relative efficacy and safety. To
simplify, a hypothetical malaria patient randomized into a clinical trial
for a new treatment and allocated to the control group could potentially
receive the standard of care of chloroquine; a malaria patient allocated
to the experimental group would be given the experimental drug. Al-
ternatively, if ethical considerations allow it and withholding a stan-
dard of care treatment for the duration of the trial is not considered
potentially damaging to the person’s health, participants in the control

group may also be “treated” with a placebo. In drug trials, this pill
does not contain an active ingredient (a “sugar pill”).
It should be noted, however, that the standard of care is hard to de-
fine for certain conditions that are usually the target of DTx (e.g., be-
havioral conditions). This may require additional considerations when
the process of defining the value proposition of the DTx, where stake-
holders and regulators w ill require an outcome improvement over the
standard of care. Researchers can address t hese considerations by de-
signing multiarm t rials where one group receives the DTx interven-
tion added to a given standard of care assessed against the standard
of care alone.
Another way to address these considerations is to use as a control a
digital placebo or a waitlist. This is often possible when evaluating dig-
ital interventions that address m ental health in patients that have
yet to receive treatment in a conventional setting. Suitable digital pla-
cebos are based on and delivered through an existing app framework,
whereby the actual experimental intervention is replaced with generic
information about the patient’s condition, a game, or a mock treat-
ment. A good example of a DTx solution that was developed and vali-
dated using alternative controls is EndeavorRX, formerly Project Evo,
Akili Interactive’s gamified DTx solution for the treatment of atten-
tion deficit hyperactivity disorder (ADHD). Evo harnesses a video
game–based approach with increasing levels of distractions and diffi-
culty but is carefully designed to target specific brain regions and im-
prove cognitive function. In the clinical validation stage, Evo’s devel-
opers created a digital placebo, also a video game (Boggle), where the
patient would put together words from a set of letters. This placebo
did not mediate effects on the ADHD condition of the patients while
requiring the same duration of screen time as the Evo intervention.
Sleepio, a DTx solution for the treatment of insomnia via digital cog-
nitive behavioral therapy, is another digital health technology that
created a digital placebo using the original platform and design but
without the active therapeutic ingredient. A point to consider regard-
ing the use of a digital placebo is controlling for any other factors that

could impact improvement (e.g., placebo effect, effect of engagement
with provider, or any other support that might come with the tool).
In addition to digital placebos, waitlist-based control groups receive
treatment after the study and are a suitable alternative in cases when
the participants’ condition allows for a temporary delay in the treat-
ment. W hether opting for either a digital placebo or a waitlist, how-
ever, researchers should expect an inflated effect size of the interven-
tion being studied b ecause of the effective absence of any treatment
in the inactive comparator-based control arm. As the technology ma-
tures, too, it w ill give rise to opportunities for new versions of DTx to
be tested against their predecessor or a competitor. In such cases, the
“old” DTx could be used as an active comparator, although it may still
not be recognized as the standard of care. Nonetheless, we expect that
the standard of care landscape w ill change dramatically with the in-
creasing adoption of DTx.
DTx Validation across Cultures and Stakeholders

DTx platforms able to successfully address their intended use while
backed by clinical evidence at scale will ultimately transcend the bound
aries of the country in which they w ere developed and first tested.
As DTx platforms are eventually deployed across different cultural,
national, and linguistic environments, important considerations per-
taining to acceptance, uptake, and adoption may need to be addressed.
While social differences might not be a challenge unique to DTx, they
are a more prominent issue from a technology validation perspective
than they are in the case of clinical research evaluating the effectiveness
of a new drug or a physical device (e.g., an orthopedic brace). This is
because DTx platforms target behavioral components, which may be
closely related to societal norms and can therefore vary dramatically
across cultures. DTx developers and researchers should take these dif-
ferences into account when building their platforms, designing a trial,
and mapping out implementation stages so that the outcomes reflect
the true validity of the DTx when deployed in multicultural populations.

Likewise, emphasis should be placed on recruiting trial participants
that reflect the diversity of the population in which the solution will be
deployed.
Just as DTx designers and innovators need to account for diverse
and culturally unique user bases, they should also expect variability
in their potential users’ personal attitudes toward the use of DTx. This
variability may stem from such factors as age, socioeconomic status,
familiarity with technology, or personal preference. Hence, as prom-
ising as the technology may be, it w ill only be an effective solution if
people actually use it. DTx organizations need to conduct the neces-
sary target population research that provide the initial insights about
the expected adoption of their proposal. One way to obtain this infor-
mation is by designing DTx clinical t rials that use mixed quantitative
and qualitative methods so as to assess acceptability and route toward
widespread adoptability alongside safety and efficacy.
We would like to take a moment h ere to emphasize the critical role
qualitative researchers play in DTx development. As noted earlier in
our discussion about the importance of having the right team, qualita-
tive research professionals bring an essential skill set by using an eth-
nographic lens to design questionnaires, surveys, focus groups, and
other methods that capture user engagement and identify potential
shortcomings of a digital intervention from a usability standpoint.
These qualitative research methods can also be used to assess the us-
ability, acceptability, and likely adoption of the DTx by all relevant
stakeholders (i.e., caregivers, clinicians, health insurance reimburse-
ment leads). We will see later in chapter 10, “Supporting Adoption of
DTx,” the importance of having such data points early in the DTx de-
velopment road map.
Clinical Validation Does Not Translate into Widespread Adoption

Regulatory approval alone may not be enough to convince providers,
payers, and investors of your DTx solution’s value. This is b ecause the
regulatory approval after clinical trials is based on how “capable” an

intervention is in a controlled setting, which can ultimately differ from
the real world. The measures used to reflect the performance of an in-
vestigational therapy in t hese different settings are known as efficacy
and effectiveness, respectively.
Efficacy refers to the performance of a treatment u nder the con-
trolled environment of a clinical trial, with strict criteria for the in-
clusion and exclusion of trial participants. For example, possessing cer-
tain comorbidities may preclude someone from participating in a
clinical trial while many f uture patients, who w ill eventually undergo
treatment with the intervention post-approval, may have those same
comorbidities. Nonetheless, in order to assess an intervention from a
trial perspective, the importance of inclusion and exclusion criteria is
evident. Effectiveness, on the other hand, refers to the performance
of that same treatment in the real world, where it may be prescribed
to and used by patients who did not participate in the trial and who
may have very diff erent characteristics from those who did participate.
Clinical trials conducive to market access measure efficacy, yet stake-
holders demand both efficacy and effectiveness. Of course, DTx are
subjected to clinical trials, an integral part of cementing clinical evi-
dence for premarket submissions to regulatory bodies. But unlike tra-
ditional clinical trials, DTx technologies and devices are typically
tested directly in the market, in real-world settings.
Provided that a DTx solution has received FDA marketing approval,
the company that develops and commercializes it must continue to
demonstrate its effectiveness in the post-approval period, commonly
by conducting a phase IV trial. Of note, DTx trials and their inherent
potential for decentralization may markedly expand the impact of
their resulting datasets due to the capacity for increased frequency of
longitudinal patient monitoring over traditional study designs and lo-
gistics, among other considerations.
As mentioned earlier, some countries are starting to lay the ground-
work to expand DTx by developing guidelines that can potentially en-
able such technologies to gain a foothold in the market. The National
Institute for Health and Care Excellence (NICE) in the United Kingdom,

for example, has developed an evidence standards framework for digi-
tal health technologies in collaboration with NHS E ngland, Public
Health E ngland, and MedCity, a not-for-profit cluster organization for
the health and life sciences sector in the London area. The framework
describes standards of evidence—clinical and economical—required
for a DTx to be considered valuable for the health care system. It as-
sists DTx developers to create a plan to fulfill these requirements and
helps commissioners to decide on whether to contract a DTx.7 In 2019,
Germany’s parliament passed the Digital Healthcare Act (Digitale-
Versorgung- Gesetz, also referred to as the DVG), which regulates the
reimbursement of DTx. Specifically, publicly insured patients (90% of
the German population) gained coverage for Class I or Class IIa (low
risk) software medical devices or stand-alone software purposed for
disease diagnosis, monitoring, and treatment. The DVG has also en-
abled entrepreneurs to apply for preapproval through a fast-track pro
cess. Subsequently, during a 12-month in-market trial period, the
digital solution is tested for efficacy.8 In 2020, the National Cryptologic
Center in Spain released a guide with best safety practices for health
app developers, including minimum requisites for safe data treatment
and storage.9
N-of-1 Trials
As we mentioned earlier, controlled clinical trials—and RCTs in
particular—are considered the gold standard of realizing and validat-
ing evidence-based medicine. However, many would-be trial partici-
pants are often unable to participate in studies due to inclusion and
exclusion criteria. Since trial findings are largely based on a treatment’s
population-w ide effects instead of individual outcomes, this is likely
an opportunity to further enhance treatment efficacy and even pin-
point more responders to treatment compared to outcomes from tra-
ditional trial designs. Additionally, although a lack of comprehensive
data precludes reliably calculating average RCT costs, it is indisputable
that conducting RCTs can be a very expensive process.10

By contrast, N-of-1 trials, as they are known in the clinical litera
ture, generate evidence for an individualized treatment effect. T hese
are clinical studies involving only one participant, whereby that per-
son typically undergoes control (e.g., placebo, no treatment, or active
standard-of-care treatment) and experimental intervention phases,
with adequate “washout” periods in between, defined as the time be-
tween treatment periods. The intervention sequence can range from
a simple AB format (where A = control, B = experimental) to more com-
plex sequences (e.g., ABBA, ABABAB) that allow researchers to assess
the treatment effect over time, accounting for what statisticians call
random errors, such as intermittent symptoms or temporary habits or
diets that may affect the outcome. Essentially, a longer sequence can
help to discern w hether the results are consistently due to the assigned
treatment, whereas an AB sequence may make it more difficult to know
whether they are due to the treatment or to an unrelated, confound-
ing factor.
By design, N-of-1 trials are patient-centric. With each patient effec-
tively serving as their own control while also being addressed with
multiple interventional permutations, more information may be
gleaned from each participant compared to traditional trial designs.
Optimal study targets for this type of trial are progressive or chronic
conditions that are characterized by frequent symptoms and for which
a known and actionable biomarker exists. Although data collection can
be done in multiple ways, the “bespoke” nature of individualized clin-
ical trials makes them an ideal environment for the use of sensor-based
wearables, which facilitate longitudinal, passive (i.e., without the pa-
tient needing to actively answer any questionnaires, keep a diary, re-
spond to a clinical trial nurse’s questions, submit himself or herself the
physical measurements, etc.) collection of outcome measurements, of-
ten a core component of DTx technologies.
Additionally, b
ecause of their unique study protocols designed to
suit individual participants with specific health conditions, N-of-1
trials can benefit from continuous iterative improvements to their
underlying data analytics systems. T hese systems or algorithms,

whose outputs can lead to more precise and timely diagnostics, prog-
nosis prediction, and personalized care processes, are one of the
main types of use cases for the application of artificial intelligence
(AI) in the health care domain. (Read more about AI applications in
appendix A.) By leveraging preexisting population or individual data,
as well as prospectively acquired data that are calibrated to the pa-
tient u
nder treatment, AI can uniquely advance the personalization
at the foundation of N-of-1 trials. With this information, AI can also
generate predictive models or particu lar profiles to provide person-
alized diagnosis and therapy, allowing trial designs that dynamically
customize the intervention at a participant level. CURATE.AI is an
example of how AI can be successfully integrated in such t rials.
CURATE.AI: A “Small Data” AI Approach to N-of-1 Trials

It is common knowledge that patients vary from one another. However, it is also
important to note that patients can vary from themselves during the course of
treatment. While it is also common knowledge that therapies exhibiting efficacy in a
patient at the start of treatment may lose efficacy over time, treatments that don’t
mediate efficacy at the outset can, in certain circumstances, be misperceived as
ineffective for the patient under care. If they can be administered optimally (defined
as delivering the right therapies at the right respective doses), however, they can
be effective. Resolving the right drug and dose parameters to optimize treatment
for each patient, and doing this in a longitudinal and dynamic manner can be
insurmountable using traditional population-derived approaches. This is where
CURATE.AI comes in.
Current drug dosing methods rely on institutional guidelines and experience-
based decisions by clinicians, which can lead to underdosing or overdosing events
and the associated reduced effect or toxicity, respectively. Common AI approaches
to treatment optimization train an algorithm to find statistical patterns based on
preexisting population data, and the identified pattern is then used to treat an
individual based on their characteristics. This big data approach to AI has some
challenges, however. For example, it often requires large and comprehensive
datasets that are representative of the population. Also, this approach might be

suboptimally suited to certain subpopulations with very specific health conditions
or characteristics. And it is very resource-intensive.
By contrast, CURATE.AI is a small data11 platform, operating based on a single
individual’s dataset. Using an AI-derived algorithm, CURATE.AI builds a profile for that
individual that predicts that person’s likely response to different drug doses. Using
this information, CURATE.AI can dynamically adapt the treatment dose for the most
optimal outcome for that person. CURATE.AI is indication-agnostic, meaning that it
can be applied to virtually any condition where there is a known and quantifiable
biomarker (e.g., tumor size reduction, cognitive improvement) that yields a dose-
dependent response to changes in input (e.g., drug dose, cognitive therapy
intensity).
An important advantage of CURATE.AI over traditional personalization methods is
that it does not require large amounts of population data, and the user profiles it
builds can be recalibrated in the event of physiological changes. CURATE.AI is a
clinical decision support system handled by a team of scientists that analyze the
data and provide recommendations to clinicians. It has been successfully applied
to drug dose personalization in solid tumor, infectious diseases, and posttransplant
immunosuppression trials, and it is being currently tested for its applicability to
improving cognitive function in people recovering from brain tumor surgery.12
The CURATE.AI proof-of-concept study was conducted as a prospective random-
ized controlled pilot trial where eight liver transplant patients were administered the
immunosuppressive drug tacrolimus. The posttransplant patients needed to
maintain their tacrolimus levels within an individually assigned target range that
avoided both toxicity events and transplant rejection. The researchers used
individual data generated by standard-of-care dosing methods to calibrate
patient-specific CURATE.AI profiles and then leveraged the profiles to guide patient
dosing. Patients randomized into the CURATE.AI group had less variability in
tacrolimus trough levels than standard-of-care control patients. Of note, CURATE.
AI-treated patients were discharged, on average, nearly one month earlier com-
pared to standard-of-care patients.
This pilot study exemplifies a significant series of N-of-1 where every patient
received a different treatment recommended by the same principle (CURATE.AI) and
served as an example of harnessing digital medicine to innovate both trial design
and novel capabilities in personalized treatment. Several qualitative studies are

being conducted in anticipation of the future deployment of CURATE.AI to gauge the
thoughts and attitudes of medical professionals and students toward using an AI
platform as an aide during their decision-making process. For more insights into the
importance of this process, please see chapter 10 on how to support DTx adoption
and gain clinical buy-in.
Institutional Review Boards and N-of-1 T rials

A critical step toward the initiation of an N-of-1 trial is the IRB review
process. N-of-1 trials can have two main objectives: personalized treat-
ment or generalizable knowledge. The IRB typically presents different
requirements based on the main goal. In cases where the intended effect
is to evaluate and adjust treatment in direct patient care, an IRB should
be consulted to identify the procedures and compliance mea sures
needed to initiate this form of care, as guidelines can vary substantially
between institutions and jurisdictions. More commonly, however, N-
of-1 trials aim to produce generalizable knowledge that could inform
optimal treatment for individuals with similar characteristics in the
future. In these cases, the IRB may request that the study design be
adapted to enable aggregated analysis of multiple identical N-of-1 trials.
CONSORT
To improve the reporting of RCT outcomes and to optimally commu-
nicate a trial’s design, conduct, analysis and interpretation, the CON-
SORT (Consolidated Standards of Reporting Trials) Group launched
the CONSORT 2010 Statement, an evidence-based guideline with a set
of minimum recommendations to produce high-quality, standardized
clinical evidence. Subsequently, the group developed the CONSORT ex-
tension for N-of-1 trials (CENT 2015), a framework to facilitate and
standardize the reporting of N-of-1 trials.13 The CENT 2015 guidelines
highlight methodological considerations that differ from traditional

clinical trial designs, such as statistical analysis, ancillary analysis, and
outcome reporting. This framework also opens a critically important
pathway toward realizing additional impact from N-of-1 trials. Spe-
cifically, adherence to such reporting guidelines allow for aggregated
analysis of results from similar studies to provide further generaliz-
able conclusions.
In summary, well-designed N-of-1 trials can serve as a pathway
toward practice-defining evidence for individualized treatments. As
such, they are well suited for testing DTx, which are highly customiz-
able by nature due to the immense volumes of personal data they are
designed and engineered to gather. In this context, the importance
of AI is only expected to grow as it can tap into individual behaviors
registered in a DTx, analyze the data and specific needs of a subject,
and develop a targeted and potentially dynamically modulated inter-
vention strategy.
Decentralized Clinical T rials

Digital health technologies leverage computing platforms, software,
sensors, and telecommunications for health care uses, allowing the de-
velopment of mobile health-compatible wellness and DTx apps. T hese
technologies have enormous potential to give rise to new paradigms
of remote clinical validation as well as downstream health care deliv-
ery, patient monitoring, and broader care-at-home workflows. T here
have been several denominations for these trials where mobile tech-
nology and decentralization converge: remote, virtual, siteless, digi-
tal, federated, and direct-to-patient are some of the names used by dif
ferent groups, sometimes with disparate definitions for the same
name. The most widely accepted denomination is the decentralized
clinical trial (DCT). DCTs are described as trials executed through tele-
medicine and mobile/local health care providers, using processes and
technologies differing from the traditional clinical trial model.14 Sim-
ilarly, remote decentralized clinical t rials (RDCTs) have been defined

as “an operational strategy for technology-enhanced clinical t rials that
are more accessible to [participants] by moving clinical trial activities
to more local settings.”15
DCTs utilize mobile technology and local health care providers,
eliminating the need for substantial centralized infrastructure com-
prised of staff, investigators, and vendors across potentially multiple
trial sites. T
here are many possible partial centralization options to
apply depending on patient needs and the specific characteristics of
the study; these configurations are adopted in so-called hybrid trials.
DCTs enable clinical trial teams to collect data directly from par-
ticipants, without requiring in-facility visits. Even passive data collec-
tion without any direct information of the participant may be possi
ble. T hese logistical considerations may allow for more convenient,
patient-centric trials. With home-based participation, patient engage-
ment may increase in all phases. Importantly, participant recruit-
ment numbers may be higher, and therapy adherence, where applica-
ble, may be aided for the participant. In summary, well-designed DCTs
can improve data collection, data quality, and patient monitoring com-
pared to traditional approaches. With DCTs, trial costs may be sub-
stantially reduced and timelines may be accelerated compared to tra-
ditionally conducted clinical research. Furthermore, data acquired
from scenarios that resemble care at home may mirror real-world out-
comes more effectively than clinically acquired findings, potentially
better informing eventual consumer use considerations. The scattered
territorial distribution of participants may also contribute to this ef-
fect by increasing biological and cultural diversity.
When designing DCTs or hybrid t rials, it is important to consider
trial, patient, and disease specificities to assess the optimal degree of
decentralization (or the idea of decentralizing at all). Key questions
include:
• Are t here any assessments that need to be done in a clinical site?

• What is the geographical distribution of potential trial partici-
pants (e.g, rare disease patients may be more dispersed)?

• What are the local regulations regarding obtaining informed
consent, remote interventions, and data collection?
• Is the device appropriate for independent use by patients less
used to digital technology?
Additionally, many patients regard h uman connection among the most

important aspects in the clinical trial experience.16 Researchers w ill
need to find the appropriate balance that allows patients some degree
of in-person interaction with the clinical team but also the freedom
to benefit from and contribute meaningfully to a trial remotely.
Several other challenges of DCTs are worth mentioning. In a DCT,
patient authentication (e.g., verifying that a sensor is being worn by
the participant) will be less straightforward than in a traditional trial.
Ideally, measures should be in place to minimize t hese occurrences,
and leveraging technology for authentication will be increasingly com-
mon (e.g., facial or other biometric identifiers).
Patient engagement with the intervention might, in some cases, de-
crease due to the decentralized design. As explained earlier, some in-
dividuals may prefer face-to-face interaction to feel involved in the
intervention/treatment process. A hybrid approach that meets a
broader spectrum of patient needs may in this case be an optimal and
truly patient-centric approach. However, researchers should ensure
that the data captured in different settings are comparable.
Regulatory authorities like the FDA and the European Medicines
Agency have signaled that they are supportive of decentralized t rials
and are currently developing regulatory frameworks to modernize
DCT oversight and ensure data leveraged from these trials are handled
using transparent data standards.17 Moreover, the CTTI has issued
legal, regulatory, and practical recommendations on the adoption of
decentralized clinical t rials.18 It is a deliberative process, boosted by
the appearance of COVID-19 and the disruption the pandemic has
caused in the execution of trials across the world, that w ill likely cul-
minate in the alignment of international regulatory bodies in this re
spect and a potential mainstreaming of DCTs.19

Connected Clinical Trials
Despite digital health technology advancements, clinical trials are still by and large
conducted the old-fashioned analog way. Current manual processes, such as data
tracking and supply management, are error-prone and slow, while deficient patient
monitoring and communication can increase patient risk. Adapting clinical trial
infrastructure so that it integrates new technologies would increase trial efficiency,
flexibility in the configuration of different trial aspects, and data security.
Realizing the power of digital solutions to optimize clinical trial efficiency and
compliance, Tata Consultancy Services and Janssen Research & Development
jointly built Connected Clinical Trials (CCT), a module-based platform for the
digitalization and unification of all aspects of a trial. The module’s eTracking feature
facilitates registering and following the kits that are sent to participants; eLabel
provides patient-specific electronic drug labels; eCommunication enables custom-
ized, real-time notifications; and eAdherence verifies medication intake thanks to
smart blister packs that leave a time stamp when a pill is extracted, potentially
allowing clinical trial staff-initiated intervention to improve timely dosing. Data
collected through these features will subsequently be analyzed and sent to the
relevant team members.
A phase I technical trial to assess the CCT platform was carried out with healthy
subjects to test integration, communication, and user satisfaction. Trial participants
were provided with smart blister packs containing placebo and a smartphone
where they could download the study app. The results showed strong compliance
and high acceptance of electronic labels, smartphone functions, and data upload-
ing by patients, and demonstrated the potential to eliminate or reduce the use of
paper and associated data-entry errors, improve the experience of trial staff, and
obtain real-time insights to strengthen treatment safety.
Challenges with DTx Data Analysis, Sharing, and Storage

Most clinical trial data are collected directly from the patient. When this
collection is done digitally, it is usually done passively, but digital tech-
nologies can generate much higher volumes of data in real time with
superior accuracy and frequency compared to traditional methods of

data collection. Given this avalanche of data, automated AI-based
analysis techniques, such as machine learning, can be used to monitor,
analyze, predict, and detect trends and anomalies in these large data-
sets on an individual and population level. For physicians, the rapid
identification of patterns and possible adverse events opens the door
to advancing dynamically personalized care into clinical practice, and
it can ultimately lead to improved outcomes for their patients or, if
used as a screening tool, reveal early signs of disease in the target
population. In their most advanced form, machine-learning algo-
rithms also make it possible for DTx technologies to learn by contin-
uously streaming and updating user data and adjusting outputs. This
increasing complexity of information flow, potential for automation,
and potential ability to dynamically evolve the treatment approach
represent a challenge for data analysis and regulation.
Data storage and sharing of a continually growing and changing
dataset also present challenges, as current digital infrastructures are
generally not adapted to store longitudinal data or to perform real-
time analysis. Additional data storage and security infrastructures,
data sharing agreements, and other factors surrounding compliant
stewardship may need to be considered for dynamically evolving data
repositories. Traditional electronic data capture forms are based on pa-
per processes and cannot store diff erent types of data in one place. As
a result, electronic records may end up in separate repositories, hin-
dering communication between stakeholders and systems (e.g., de-
vices, researchers, vendors, sites, CROs, trial sponsors).
It is of paramount importance to acquire tools that integrate data
management and collection from diff erent sources and in nonconven-
tional formats into a centralized resource. In recognition of this rap-
idly evolving space, regulatory agencies are starting to lay out a holis-
tic framework to assess the pertinence of a DTx platform’s access to
market based on a number of factors. T hese include intended purpose,
quality control mechanisms, level of h uman intervention, value added
to the clinical workflow, risks and mitigation measures, and continu-
ous learning plans and version controls for subsequent iterations of

the DTx, among others. At the foundation of this infrastructure, all
parties h andling data should have aligned knowledge of the trial data
flow and storage. To provide further insights into this important topic,
CTTI’s Mobile Technologies recommendations are a helpful resource
in this respect.20
Deciding on the Optimal Regulatory Route

Due to the recent inception of the DTx space, guidance from the in-
ternational regulatory community is still evolving. Generally, DTx reg-
ulatory approval or clearance requirements depend on the product
claims regarding risk, efficacy, and intended use. Validation of efficacy
and safety claims by regulatory or equivalent national bodies is re-
quired for DTx that treat or manage a disease, whereas for DTx plat-
forms that claim to improve health function or prevent a disease, the
degree of oversight may depend on additional f actors outlined by lo-
cal regulatory frameworks.21
In the United States, DTx platforms are regulated by the software
as a medical device (SaMD) framework developed by the International
Medical Device Regulators Forum. Most of the platforms have been
designated as Class II (moderate to high risk) or Class III (high risk)
devices as per FDA classification. Market clearance can be obtained
with a 510(k) premarket submission that demonstrates the device is
substantially equivalent to a legally marketed device (called a predi-
cate). This clearance—the usual route for Class II devices—avoids the
need of a costly and lengthy clinical trial, as long as the sponsor, de-
fined as the person or organization who takes responsibility for and
initiates the clinical investigation, ensures compliance with applica-
ble risk controls.22
If the FDA designates a device as not substantially equivalent to the
predicate, or if t here is no predicate, or the sponsor suspects the de-
vice would receive a Class III classification, a De Novo risk-based clas-
sification can be requested, which can assign to the device a lower risk
classification. The De Novo route allows for the marketing of devices

for which applicable risk controls provide reasonable assurance of
safety and effectiveness for the intended use.23
Class III devices require premarket approval (PMA),24 and to collect
clinical data for the PMA, an investigational device exemption is re-
quired before using the device in a clinical trial and shipping it across
state lines.25 When possible, deciding between FDA clearance and ap-
proval should take into account what kind of evidence w ill be required
by the sponsor’s target stakeholders.
It is important to note that the same product can be subject to a dif
ferent regulatory workflow, including no regulation at all, depending
on the intended use claimed by the sponsor (e.g., wellness vs. diagno-
sis). In some cases, the sponsor w ill claim their product is a medical
device, and the FDA w ill determine that the product is actually a de-
vice but w ill exercise “enforcement discretion” and not regulate it.26
The FDA could also grant temporary market authorization for a DTx
during the course of a public health emergency after an interactive re-
view process culminates in an Emergency Use Authorization.
In the European Union, DTx platforms qualify as medical device
software (MDSW) and are covered by the EU Medical Device Regula-
tion, or Regulation (EU) 2017/745 (EU MDR).27 MDSW is software
intended to be used, alone or in combination, for a purpose as speci-
fied in the definition of medical device in the EU MDR. The EU MDR
distinguishes Class I, IIa, IIb, and III devices, from low to high risk
and low to high regulatory requirements. Rule 11 of the EU MDR
states that software intended to provide information used to make
decisions with diagnosis or therapeutic purposes belongs to Class IIa,
except if those decisions may cause death or an irreversible health de-
terioration (Class III) or a serious health deterioration or a surgical
intervention (Class IIb). Software intended to monitor physiological
processes is classified as Class IIa, except for monitoring of vital
physiological parameters, where parameter variation could cause im-
mediate danger to the patient, in which case it is classified as Class IIb.
Rule 22 of the EU MDR applies to active therapeutic software with an
integrated diagnostic function that determines patient manage-

ment and leads to Class III.28 Consequently, DTx w ill fall into Class
IIa or higher and w ill require a Certificate of Conformity by a Notified
Body.29
In Singapore, the import, manufacture, export, and supply of med-
ical devices is regulated by the Health Sciences Authority (HSA). In
2021, the agency issued a guidance document on the regulatory re-
quirements of any software that falls under the medical device defini-
tion according to the Health Products Act. DTx sponsors must submit
to HSA a premarket product registration application to prove they
comply with the essential principles for safety and performance of
medical devices. Clinical evidence requirements w ill depend on the
purpose of the generated information (to treat or diagnose, to drive
clinical management, or to inform clinical management) and the seri-
ousness of the medical condition.
The fast-changing landscape of digital technologies is driving au-
thorities to develop specific regulatory pathways. The FDA launched
the Digital Health Software Precertification Program to streamline
DTx submissions from manufacturers with accredited quality stan-
dards.30 Also, the FDA’s Center for Devices and Radiological Health
released a product lifecycle–based regulatory framework for AI-and
machine-learning-based SaMDs.31 In the UK, as mentioned e arlier, the
NICE has developed an evidence standards framework for digital
health technologies to help developers understand the clinical and eco
nomical evidence standards required for their product to be consid-
ered valuable for the health care system. For the latest regulations, DTx
organizations should investigate their market-specific regulatory stan-
dards and requirements.
Conclusion
Developing a DTx solution requires the ability to prove its efficacy
through clinical validation, which is a complex process. It requires the
interdisciplinary efforts of engineers, ethicists, clinicians, and health
care decision-makers to address DTx-specific challenges arising from

the novelty and part icu lar characteristics of digital interventions.
T hose same features, however, may allow for the downstream design
of innovative, participant-centric t rials where DTx are tested with
unprecedented efficiency and objectivity. Regulatory bodies, agencies,
and associations are addressing those specificities and paving the way
for a more streamlined DTx validation process by creating guidelines,
collaborative databases, and distinct pathways for approval.
Key Takeaways
• Clinical validation is a necessary step toward DTx market introduction,
whereby safety and efficacy are assessed through established controlled
trial design principles, often a randomized clinical trial (RCT).
• DTx clinical t rials may employ digital endpoints (safety and efficacy assess-
ment criteria that are digitally collected and analyzed) that increase the
accuracy and objectivity of the final analysis, but which also need to be
validated. The Digital Medicine (DiMe) website includes a collaboratively
sourced database of validated digital endpoints.
• DTx trials can circumvent the absence of consensual standard of care with
multiarm designs testing standard of care, standard of care plus DTx, and
DTx alone.
• It is important for DTx platforms to be validated across cultures, demograph-
ics, and stakeholders through mixed quantitative-qualitative methods in
order to realize broadly actionable insights into effectiveness and safety.
• DTx are well suited for N-of-1 trials, which provide the highest-grade evidence
for individualized treatment effect.
• DTx t rials may be easily adapted to decentralized and hybrid trial designs,
which can improve recruitment, retention, and therapy adherence while
reducing costs.
• DTx platforms can generate vast amounts of trial data that can in turn lead to
substantially enhanced analytical insights compared to conventional trial
designs, provided the right infrastructure is available.
• In the context of a digitally delivered therapeutic, real-world evidence can be
equally or more relevant than controlled clinical trial outcomes.

Notes
1. Institute for Clinical and Economic Review, “Cost-Effectiveness, the QALY,
and the evLYG.”
2. ISPOR, “About HEOR,” accessed September 10, 2021, https://w ww.ispor.org
/ heor-resources/about-heor.
3. Digital Medicine Society, “DiMe’s Library of Digital Endpoints, accessed
September 10, 2021, https://w ww.d imesociety.org/index.php/ k nowledge-center
/ library-of-d igital-endpoints.
4. UCB Biopharma, “Study to Evaluate the Impact of Using Wearable Devices
in Addition to Standard Clinical Practice on Parkinson’s Subject Symptoms
Management.”
5. Clinical Trials Transformation Initiative, Developing Novel Endpoints Generated
by Digital Health Technology for Use in Clinical Trials.
6. Standard of care is “treatment that is accepted by medical experts as a proper
treatment for a certain type of disease and that is widely used by healthcare
professionals. Also called best practice, standard medical care, and standard therapy”
(National Cancer Institute, NCI Dictionary of Cancer Terms, accessed September 10,
2021, https://w ww.cancer.gov/publications/dictionaries/cancer-terms/def/standard
-of-care).
7. National Institute for Health and Care Excellence, Evidence Standards
Framework for Digital Health Technologies.
8. Federal Ministry of Health, “Driving the Digital Transformation of Germa-
ny’s Healthcare System for the Good of Patients: The Act to Improve Healthcare
Provision through Digitalisation and Innovation (Digital Healthcare Act—DVG).”
9. Centro Criptológico Nacional [National Cryptologic Center], “Requisitos de
seguridad para aplicaciones de cibersalud [Security Requirements for eHealth
Applications].”
10. Speich et al., “Systematic Review on Costs and Resource Use of Randomized
Clinical Trials Shows a Lack of Transparent and Comprehensive Data.”
11. Small data are data of accessible volume and format for h umans. The
accepted definition says that “small data connects people with timely, meaningful
insights (derived from big data and/or ‘local’ sources), organized and packaged—
often visually—to be accessible, understandable, and actionable for everyday
tasks” (Allen Bonde, “Defining Small Data,” Small Data Group (blog), October 18,
2013, https://smalldatagroup.com/2013/10/18/defining-small-data). Although
humanity has based most of its advancements on small data, the concept appeared
recently in opposition to the ubiquitous big data, which are data too large or complex
to understand or analyze by common means. Small data are cheaper, all around us,
and can have many applications, especially in the health care and marketing fields
(e.g., streamlining shopping, personalizing fitness routines, tracking habits,
predicting behaviors).
12. Zarrinpar et al., “Individualizing Liver Transplant Immunosuppression Using
a Phenotypic Personalized Medicine Platform”; Shen et al., “Harnessing Artificial
Intelligence to Optimize Long-Term Maintenance Dosing for Antiretroviral-Naive

Adults with HIV-1 Infection”; Pantuck et al., “Modulating BET Bromodomain
Inhibitor ZEN-3694 and Enzalutamide Combination Dosing in a Metastatic Prostate
Cancer Patient Using CURATE.AI, an Artificial Intelligence Platform”; Kee et al.,
“Harnessing CURATE.AI as a Digital Therapeutics Platform by Identifying N-of-1
Learning Trajectory Profiles.”
13. CONSORT, “N-of-1 T rials,” accessed September 10, 2021, http://w ww.consort
-statement.org/extensions?ContentWidgetId=47627.
14. Clinical Trials Transformation Initiative, CTTI Recommendations: Decentral-
ized Clinical Trials.
15. Trials@Home, “About RDCT’s,” accessed April 10, 2023, https://trialsathome
.com/about-rdcts.
16. Anderson, Borfitz, and Getz, “Global Public Attitudes about Clinical Research
and Patient Experiences with Clinical Trials.”
17. Gottlieb, “Breaking Down Barriers between Clinical T rials and Clinical Care.”
18. Clinical Trials Transformation Initiative, CTTI Recommendations.
19. Byrne, “DIA 2021 Set to Shed Some Light on Decentralized Trials Process,
Accelerated Approval Pathways.”
20. Clinical Trials Transformation Initiative, Digital Health T rials: Recommenda-
tions for Delivering an Investigational Product.
21. Digital Therapeutics Alliance, “Understanding DTx.”
22. US Food and Drug Administration, “Premarket Notification 510(k).”
23. US Food and Drug Administration, “De Novo Classification Request.”
24. US Food and Drug Administration, “Premarket Approval (PMA).”
25. US Food and Drug Administration, “Investigational Device Exemption (IDE).”
26. Coravos et al., “Digital Medicine: A Primer on Measurement.”
27. See https://eumdr.com for the latest on the European Union Medical Device
Regulation.
28. EUR-L ex, Document 02017R0745-20170505, May 5, 2017, https://eur-lex
.europa.eu/eli/reg/2017/745/2017- 05- 05.
29. European Commission, “Internal Market, Industry, Entrepreneurship, and
SMEs: Conformity Assessment.”
30. US Food and Drug Administration, “Digital Health Software Precertification
(Pre-Cert) Pilot Program.”
31. US Food and Drug Administration, “Artificial Intelligence and Machine
Learning in Software as a Medical Device.”

• Chapter 8
F actors That Influence the Implementation

of New Technologies
This chapter was cowritten with Dr. Robyn Mildon, Founding Executive
Director of the Centre for Evidence and Implementation with offices
in Australia, Singapore, and the United Kingdom. Dr. Mildon is an
internationally recognized leader in the fields of implementation
science, evidence synthesis, and knowledge translation, as well as
policy evaluations in health, education, and h uman services.
DTx platforms are designed to induce cognitive, behavioral, and phys-

iological changes. As such, their effectiveness can vary dramatically
across clinics, demographic segments, and countries due to culture, re-
ligion, values, or social context. To add to this complexity, technolog-
ical, legal, and organizational factors related to the wider ecosystem
into which DTx are deployed may also influence their implementation.
These factors greatly condition local adoption of a DTx and must there-
fore be carefully identified, analyzed, and considered both in the ini-
tial design and the subsequent iterations of the technology.
Numerous research studies have been done to characterize these
factors. Resulting theories and models of technology adoption have been
proposed that can help innovators predict the response of end users.
154

hese include the theory of diffusion of innovation, the unified theory
T
of acceptance, the task-technology fit model, the use of technology
model, and the technology acceptance model.1 These models define key
factors that influence the adoption and use of technology, such as per-
ceived ease of use of a technology, perceived usefulness, social impact,
facilitating conditions, attitudes to use, and users’ behavior.2
As the evidence on f actors specifically and uniquely relevant to DTx
adoption is still emerging, DTx organizations should learn from sim-
ilar technologies being implemented in similar contexts. For example,
e-health technologies, electronic medical records, and clinical decision
support systems can serve as sources of learning.3 In a recent system-
atic analysis of 44 reviews of e-health technology implementation, the
authors identified 10 key f actors that influenced successful implemen-
tation (table 8.1).
Table 8.1. Factors that influence the implementation of e-health technologies
Innovation • Adaptability (of the technology to fit the local context)

characteristics • Complexity (e.g., difficulty level of the hardware and software, slow system
performance, connectivity issues, etc.)
• Cost (of the technologies and those associated with their implementation)
Outer setting1 • External policy and incentives (absence or inadequacy of legislation and
policies and liability concerns)
Inner setting2 • Implementation climate (compatibility or general fit between the technology
intervention and the organization)
Individual • Knowledge and beliefs (attitudes of practitioners t oward the technology and
characteristics beliefs about its potential benefits)
• Other personal attributes (health care professionals’ computer skills, abilities,
and experience)
Process • Planning (for implementation)
• Engaging (i.e., designation of champions and engagement of key stakeholders)
• Reflecting and evaluating (to ensure system benefits, to increase health
professionals’ acceptance through demonstration of benefits, and to secure
ongoing funding)
Source: Jamie et al., “Factors That Influence the Implementation of e-H ealth.”
1
Outer setting refers to the economic, politic al, and social contexts where the organization
resides (e.g., policies, financing, stakeholder relationships).
2
Inner setting refers to the structural, politic al, and cultural contexts where the implemen-
tation takes place.
Factors That Influence Implementatio 155

In another systematic review of studies looking at factors impacting
clinicians’ adoption of mHealth tools, as many as 55 technological, so-
cial, and organizational factors were identified.4 This provides a sense
of the complexity of implementing new technologies in health care.
Usability as a F actor of Influence

Let’s consider usability (often associated with complexity) as a factor
identified across many of t hese reviews to see how well existing digi-
tal health tools for diabetes management have integrated them into
their design. We selected this therapeutic area as an example b ecause
usability has been shown to be an adoption barrier for patients with
diabetes and physicians who treat such patients.5
Usability measures how well a specific user in a specific context can
use a solution to achieve a defined goal effectively, efficiently, and sat-
isfactorily. Conceptually, it is related to ease of use, which is the effort
required to use a particular product, serv ice, or technology. In a sys-
tematic literature review of the social, organizational, and technolog-
ical factors impacting clinicians’ adoption of mobile health, usability
was identified as one of the crucial elements impacting one’s intention
to use such technologies, but usability alone was not enough for user
acceptance.6 Layout, interface, and culturally appropriate and patient-
centered design were also seen as key influencing factors in several
studies included in the review.
In another systematic review, when 66 diabetes mobile applications
with usability as an evaluation f actor w ere compared and graded on a
5-point scale, the average score was 3.3, with scores varying from 2.5 to
4.1. This assessment reflects a neutral (or average) expression of these
solutions’ usability. It was also found that only 8% of mobile applica-
tions for diabetes w ere sufficiently interoperable to enable connection
to an external sensor or measuring device, such as a blood glucose
monitor.7 For innovation and business leaders focused on providing a
seamless, satisfying user experience, a technology that leaves room for
improvement in terms of usability will encounter challenges in being

a dopted. Patient-centered solutions, leading to better patient experi-
ence and therefore higher levels of user satisfaction, can result in bet-
ter patient engagement. Therefore, it w ill be essential for DTx organ
izations to not only define what would be a “good enough” patient
experience while using the DTx but also consider how this experience
can catalyze sustained user engagement while ultimately facilitating
widespread adoption.
A DTx solution through its design, features, and overall usability
should connect with its users—patients, caregivers, and clinicians—
on a functional and potentially emotional level as well. For example,
more research and consideration are needed surrounding the chal-
lenges and attitudes elderly patients or individuals with cognitive
impairment have with maneuvering digital apps. Insufficiently ad-
dressing these challenges may reduce the usability of the technology
for t hose segments of the population. Therefore, it is critical to un-
derstand the needs and motivations of the target users of a given
DTx solution in the context in which they intend to use it, as well as
the interactions with the app that w ill be required of them to fulfill
those needs.
Local Context Matters

To optimize the pathway to widespread adoption of a DTx platform,
usability and other factors that influence the implementation of digi-
tal health technologies should be user-centric. We stress the impor-
tance for DTx innovators to deeply engage with the contextual phe-
nomena that impact and interact with the heterogeneous needs at the
individual user level, w
hether that be for patients, caregivers, or phy-
sicians, as well as the ability to answer the specificities of a local con-
text. This is not a trivial task. Even established, multinational com-
panies undertake substantial effort to understand how contextual
factors impact technology implementation and adoption. These efforts
can lead to important lessons learned and a wealth of knowledge that
can help f uture entrepreneurs.

In 2018, Google deployed a deep learning program across 11 clinics
throughout Thailand to screen for diabetic retinopathy (DR) and
evaluate the performance of the algorithm in a real-world clinical set-
ting. The system sought to improve clinical workflows and patient out-
comes in a country with a shortage of ophthalmologists and where
diagnosis and treatment of diabetic retinopathy can be strengthened
by a more efficient health care serv ice. The program established the
importance of accounting for several socioenvironmental factors at the
foundation of their solution for the local context, which w ere key
drivers of technology performance and the clinical experience for both
the patient and clinical team.8 Let’s explore in detail the four f actors
initially deemed to be critical to the success of Google’s AI system and
that have served as a valuable source of insight into f actors that drive
patient and provider adoption.
The Importance of Localization, Learnings from Google

With regard to technical per formance, Google’s AI system was
impressive—it could detect referable cases of retinopathy with an ac-
curacy of 90% using only the analysis of a single photo of a diabetic’s
retina.9 The value proposition was simple—provide ophthalmologist-
grade diagnostic capability at scale. The screenings didn’t require
ophthalmologists to be involved at all. The nurses simply needed to
take photos of the retina and upload them into the Google’s AI sys-
tem, and within approximately 10 minutes the patient could expect a
diagnosis. By comparison, the traditional process required sending
the photos to ophthalmologists and took from 2 to 10 weeks to get
back.
Google’s AI-based screening system was a promising solution
to the common problem of health care workforce shortages in
low-and middle-income countries. Its deployment illuminated
several key factors that are critical toward clinical utility and user
adoption.

actor #1. Presence of Consistent Workflow and Screening
F
Conditions across Clinics
In an ideal health care system, clinical workflows and patient manage-
ment protocols are well documented, consistent from clinic to clinic.
During the deployment process, it was noted that there was a high de-
gree of variation in the eye-screening process from clinic to clinic.
This was driven by the large degree of autonomy given to the nurses
to organize this process depending on resources available at the time
and basic infrastructures at their disposition (e.g., only two clinics had
a dedicated screening room).
As a result of this variability, approximately 80% of the submitted
images fulfilled the AI model’s quality requirements, which prolonged
the nursing team workflow. The image variability resulted in the re-
take and resubmission of images, adding 2–4 minutes spent per pa-
tient while local protocols allocated 90 seconds per patient due to the
high volume of patients requiring screening. This consideration re-
vealed the importance of aligning technology execution with local
workflow considerations.
Factor #2. Presence of a Reliable Network Connection

In a health care setting with a strong internet connection, image pro
cessing and display typically requires a few seconds. However, due to
the local IT infrastructure, Google experienced slower and less reliable
connections than anticipated. Some images took up to 60 seconds or
even 90 seconds to upload, which slowed down the photo analysis pro
cess and lowered the number of patients that could be screened com-
pared to original estimates.
Factor #3. Patients Embracing the Proposed Solution

When the technology and internet connection worked well, patients
would have their diabetic retinopathy result within 10 minutes. Fol-
lowing a DR-positive diagnosis, and if acceptable images could not be
obtained a fter several tries, the study protocol recommended that the
nurse advise patients to see a specialist. In Thailand, however, a referral

to a nearby hospital could mean driving up to one hour, repeated
queuing, and additional costs to the patient, which represented a sub-
stantial burden for the patients. In one clinic, up to 50% of the pa-
tients elected not to join the study as they w ere not agreeable with
this potential ophthalmologist referral, and the associated “risk” to
travel. In this case, despite access to a powerful technology platform,
the workflow and potential for additional referrals created a challenge
for widespread adoption.
Factor #4. Nurses Embracing the Proposed Solution

A key aim of the AI-based retinopathy screening system was to opti-
mize clinical workflows and reduce the nurses’ workload. The AI
screening system eliminated the need for the nurses to perform an ini-
tial grading of the severity of retinopathy and check for apparent ab-
normalities. In practice, while the solution shortened the workflow by
one step, additional steps were added, and infrastructural challenges
were illuminated during the trial process. Nonetheless, the Google team
recognized that “formative research that provides a strong understand-
ing of clinical users and their context is critically important to the suc-
cess of such a system. By incorporating human-centered evaluations
into deep-learning model evaluations, and studying model performance
on live data generated at the clinical site, we can reduce the risk that
deep learning systems w ill fail in the wild, and increase the likelihood
for meaningful improvements to patients and clinicians.”10
Conclusion
Google’s study not only provided many important insights to their
team but also serves as an example to the DTx community of some of
key factors to consider when designing their implementation strate-
gies. The success of health-related technologies does not rest solely on
their technical accuracy; it also rests on their ability to fit into the
deployment context to ultimately improve patient care.11

To ultimately change practice in medicine, any technology, including
DTx, should be designed and implemented sustainably. Understanding
the needs, attitudes, and behaviors of clinicians, health care consum-
ers, family members, and policymakers (the latter because they influ-
ence the design of health systems in which interventions take place) in
the context in which they work and live are key f actors for the intelli-
gent design and long-term adoption of DTx. In this process, DTx organ
izations can draw inspiration from mHealth tools and similar technol-
ogies to leverage their lessons learned on technology implementation
in health care. Considerations include adopting a cocreation approach
that engages all stakeholders early on. In the next chapter, we w ill take
a deep dive into understanding clinicians’ perspectives and using that
understanding to drive the adoption and prescribing of DTx.
Notes
1. Garavand et al., “Factors Influencing the Adoption of Health Information
Technologies.”
2. Garavand et al., “Factors Influencing the Adoption of Health Information
Technologies.”
3. Jacob, Sanchez-Vazquez, and Ivory, “Social, Organizational, and Technologi-
cal F
actors Impacting Clinicians’ Adoption of Mobile Health Tools.”
4. mHealth is a subcategory of e-health defined as “medical and public health
practice supported by mobile devices, such as mobile phones, patient monitoring
devices, personal digital assistants, and other wireless devices” (Jacob, Sanchez-
Vazquez, and Ivory, “Social, Organizational, and Technological Factors Impacting
Clinicians’ Adoption of Mobile Health Tools”).
5. Klonoff and Kerr, “Overcoming Barriers to Adoption of Digital Health Tools
for Diabetes.”
cal F
7. Arnhold, Quade, and Kirch, “Mobile Applications for Diabetics.”
8. Beede et al., “A Human-Centered Evaluation of a Deep Learning System
Deployed in Clinics for the Detection of Diabetic Retinopathy.”
9. Beede et al., “A Human-Centered Evaluation of a Deep Learning System.”
10. Beede et al., “A Human-Centered Evaluation of a Deep Learning System,” 10.
11. Shah, Milstein, and Bagley, “Making Machine Learning Models Clinically
Useful.”

• Chapter 9
Recommendations for DTx Implementation
The section in this chapter on addressing implementation traps was

cowritten with Dr. Robyn Mildon, Founding Executive Director of the
Centre for Evidence and Implementation. Dr. Mildon is an internationally
recognized leader in the fields of implementation science, evidence
synthesis, and knowledge translation, as well as in program and policy
evaluations in health, education, and human serv ices.
Following the identification of factors that may influence DTx imple-

mentation in health care settings, DTx inventors are approaching the
implementation stage. At this point, health care organizations w ill
carefully assess the potential of new technologies, including DTx,
against the risk they pose to patients, physicians, clinical workflows,
and the system’s business model.
Medical technology innovation holds immense potential, but there
may also be health risks requiring consideration. That is why r unning
a risk-benefit analysis of adopting new technologies underpins much
of the decision-making processes that occur in health care. Due to the
stakes involved, however, the default attitude at the point of initial
162

assessment is typically one of strong aversion to risk. While this may
be perceived as a frustrating hurdle from the perspective of an emerg-
ing DTx startup, by understanding the responsibility health care sys-
tem managers have in such evaluations and embedding in the design
of the DTx the risk management measures that anticipate elevated
scrutiny, the chances that it passes the initial due diligence steps w
ill
significantly improve.
For this reason, in this chapter we highlight the expectations of
health care system managers when it comes to implementing DTx at
their hospital or clinic. We also discuss some of the most common im-
plementation pitfalls at this stage and how to avoid them with an ap-
propriate implementation strategy.
What Health Care System Managers Expect

We have already discussed the importance of demonstrating clinical
evidence that DTx technology works, as well as the other benefits that
a DTx can provide to patients, caregivers, physicians, and the health
system as a whole. As a result, we will not return to those critical points
but instead focus on the perceived risks and must-haves for DTx plat-
forms. For this segment, we spoke with Dr. Ngiam Kee Yuan, Group
Chief Technology Officer of the Singapore National University Health
System (NUHS),1 a health care system decision-maker involved with
the evaluation of digital technologies for potential integration. In our
discussion, he mentioned four elements for evaluation when consid-
ering DTx for NUHS, serving as key insights for DTx innovators:
Data Provenance and Clinical Validation

In assessing a new DTx proposition, one of the key features health care
system managers first look at is the provenance of the underlying
data (see chapter 6). To assure that a solution is safe and effective for
use by clinicians and patients, it is important for DTx innovators to
Recommendations for DTx Implementation 163

demonstrate that the data sources at its foundation have been appro-
priately vetted. Specifically, the type and quality of the data that have
been used to feed and train the tool, the context in which t hese data
have been collected, and the applicability of the model to the setting
where the technology might be deployed are essential to determining
whether it warrants further evaluation.
Security and Privacy

Privacy and security of health-related data are a top-of-mind concern
for health care system managers, and for good reason. In 2019, the
American Medical Collection Agency filed for Chapter 11 protection
after revealing it had suffered an eight-month data breach that affected
up to 20 million patients and in which the stolen data had been ad-
vertised for sale in underground web forums. In addition, among the
largest ever data breaches occurred when the records containing per-
sonally identifiable information of about 80 million customers and em-
ployees of Anthem Inc., the second-biggest health insurer in the
United States, w ere infringed.2 Hackers have also been known to in-
terfere with remotely accessible implantable medical devices, insulin
pumps, and other medical technologies. As the implementation of DTx
platforms into health care systems will simultaneously involve the de-
ployment of companion software, security against potential hackers
will be a critically important consideration for health care system man
agers. Therefore, it is essential to be prepared to answer how the
patient-generated data elicited and channeled by the DTx w ill be han-
dled, stored, and shared.
There are several approaches to protecting DTx-collected data your
technology should be designed to handle. They are centered around
detecting, separating, and securely storing personal identifiers, com-
partmentalizing highly dimensional data, encrypting data transfer,
and safely storing the entire resulting datasets. To inspire confidence
on the part of health care system managers and users, one w ill have to

incorporate the highest privacy and security standards and readiness
to adopt the health care system’s existing data governance standards.
Technology Interoperability
Integrating DTx into day-to-day clinical care requires thinking across
multiple dimensions by understanding the perspective of each of the
stakeholders involved: physicians, patients, and health care adminis-
trators. While physicians will rightfully scrutinize the clinical evidence
behind the DTx and patients its effectiveness and ease-of-use, admin-
istrators w ill be particularly concerned about its deployment within
existing infrastructure. The end goal is making care delivery more
seamless (and safe), rather than adding a new layer of complexity to
existing solutions that all too often are already working in silos. Stan-
dardized billing and product codes will be necessary for electronic pre-
scribing procedures, in addition to security requirements. Overall,
modifications in electronic health record (EHR) and workflow integra-
tion are complex, depending on how DTx w ill be reimbursed through
the different distribution channels.
Continuous Evolution of DTx

The continuous integration, continuous delivery, and continuous de-
ployment (referred to as CI/CD in technical shorthand) of DTx tech-
nologies is critical to their utility. These concepts refer to how DTx plat-
forms learn from the data and feedback they collect over time in
order to improve health outcomes through high-precision insights and
by providing a seamless experience to clinicians and patients. They
also refer to how DTx adapt to potential changes in the system within
which they have been deployed. This is applicable to both software and
hardware, such as being compatible with new versions or updates to a
hospital’s EHR system or to a patient’s mobile device. This further ex-
tends to how DTx fit within new mobile devices and new functionalities

as they are introduced, such as better cameras, improved processing
systems, and 5G networks.
To help DTx innovators think through t hese f actors, we invite them

to consider the following data management and security checklist,
drawn up in the form of questions. It is not exhaustive, as there are
sure to be other important issues that come up during the technology
development and implementation cycle, but it contains the essential
questions entrepreneurs should be able to address in order for their
product to have a chance of interfacing with an existing health care
system:
❒ Is the way the patient-generated data are used, stored, and

shared compliant with local regulatory, environmental, and
security standards as well as compliant with the partnering
health care organization’s standards?
❒ Do you have best-in-class data governance, privacy, and
security standards with a comprehensive risk mitigation
framework, encompassing preventable security and privacy
breaches, cybersecurity risks, and potential disruption of
care?
❒ Is there ease of DTx deployment (i.e., integration and interop-
erability) with the partnering health care organization?
❒ Is your DTx’s strategy not only safe but also scalable?
Risk concerns can be a hindrance to the adoption of DTx. Therefore,

as regulation of DTx evolves over the coming months and years, with
new regulatory pathways, governance structures, data infrastructures,
and security protocols being established, it is necessary for DTx entre-
preneurs to build a sound data stewardship and interfacing founda-
tion for their technology. This means investing in data management
and security on the one hand, and engaging early on with health care
institutions’ chief technology and information officers on the other to
make sure any new digital technology is compatible with existing and
planned systems and workflows.

Although a solid technical and safety foundation for DTx technol-
ogy is important, it does not guarantee commercial success. To prevent
DTx venture failure when moving from development to commercial-
ization, t here are a few implementation traps that DTx innovators
should be aware of and must overcome if encountered.
Addressing the Implementation Traps

Many of the experts interviewed for this book converge with findings
from the existing literature on digital health technologies on the view
that t here are four main reasons why some DTx ventures fail when
moving from development to commercialization.
No implementation strategy. Implementation or integration of DTx
in business-as-usual clinical practice should be regarded as a process,
not an event. As such, it should be planned and executed methodically
in stages rather than as a one-off launch (a concept some DTx entre-
preneurs might be tempted to borrow from the pharmaceutical indus-
try’s marketing approach to introducing new drugs). Many DTx start-
ups do not have an implementation strategy for their product and rely
on the idea that they can just “roll it out.”
Implementation complexity and replicability. T here is no simple or
universally replicable way of implementing change—such as a new DTx
platform—at scale in a deeply complex environment such as health
care. You have already witnessed how a technology, pathway, or an
approach can perform well in one setting but poorly or not at all in a
different context. Well-thought-out implementation strategies, there-
fore, typically have multiple components and are adaptable to local
health contexts, which are inherently complex because of their mul-
tiple interacting levels (e.g., patients, providers, teams, serv ice units)
and vary widely from setting to setting.3 It is therefore crucial to
understand—both qualitatively and quantitatively—the barriers and
facilitators to implementing DTx in different institutional, cultural,
and economic environments, since these w ill understandably impact
their adoption by individuals and organizations.4 In short, local

context m atters, but it also limits the replicability of technology
implementation.
Not having the right know-how. No matter how carefully an imple-
mentation strategy is designed and planned in advance, the experi-
ence of a ctual implementation often generates novel or even unex-
pected insights that can improve the strategy “on the go” and in real
time. Yet not all DTx startups have the necessary engine with the
right team and know-how, training, or (importantly) adaptive mind-
set to use what is known in business innovation as the rapid cycle
change model, an evolving implementation strategy to drive the
spread of an innovation with some potential to adapt to different
contexts.
All of t hese points are central to the implementation science (IS) ap-
proach that we introduced in our Part III introduction. To avoid fall-
ing victim to these implementation traps, DTx designers must think
flexibly, understand local contexts, use qualitative methods to explore
accepted processes and mechanisms, and be ready to adapt their in-
tervention to achieve the best fit within different settings.
When developing, deploying, and evaluating a solution across mul-
tiple sites, we recommend that innovators have an implementation
strategy based on a structured and phased approach. In short, follow
an IS approach. It starts with having a clear definition of the proposed
intervention and a plan for optimizing the components it consists of
and the f actors that influence its success. Ideally, that plan should be
aligned with the evidence based on behavior change and tested through
a series of small-scale trials in selected settings or through a rapid cy-
cle test of the envisioned change model—in this case related to ef-
fecting behavioral change (figure 9.1).
The central element of the rapid cycle testing approach is the plan-
do-study-act (PDSA) scientific method for testing a change by plan-
ning it, trying it, observing the results, and acting on what is learned.
It ensures that ideas for change are tested and adapted for local con-
texts while paying explicit attention to opportunities for iterative
improvement.

Specific
Ideas, explanations,
improvement
hunches, theories
intervention
P
A D Very small-scale test of change
using plan-do-study-act (PDSA) cycle
S
P
A D Follow-up tests and
iterative refinement
S
P Small-scale tests of change

in multiple sites
A D
S
P Widespread
implementation Changes that result in
A D
with local improvement at scale
S tailoring
Figure 9.1. Rapid cycle test of a change model of spread used in implementation
science, drawing on insights and a previous diagram in a review by Pierre Barker.
Sources: Greenhalgh and Papoutsi, “Spreading and Scaling Up Innovation and Improvement;
Barker, Reid, and Schall, “A Framework for Scaling Up Health Interventions: Lessons from
Large-Scale Improvement Initiatives in Africa”
Finally, we would like to offer a few recommendations adapted from

a systematic review of the implementation of e-health technologies5
and mobile health tools.6 T here is gap between evidence and the prac-
tical aspects of integrating and implementing DTx into real-world
clinical settings. Therefore, while evidence about implementing DTx
are emerging, e-health technologies and mobile health tools, sharing

Important
A systematic effort to deploy an intervention on a limited scale but across different
settings will help DTx organizations identify and deal with barriers that get in the
way of the implementation effort and take advantage of facilitators that support it.
The objective is ultimately to understand from an implementation standpoint not
only the type of intervention that works to drive adoption but also answer the
question on what works where and under what conditions.
Going Further with the NASSS Framework

To further your knowledge and understanding of implementation frameworks for
new technologies, we recommend familiarizing yourself with the comprehensive
nonadoption, abandonment, spread, scale-up, and sustainability (NASSS) frame-
work. It can be used to inform the design of a new technology; identify solutions
that have a limited chance of achieving sustained adoption; plan the implementa-
tion, scale-up, or rollout of a technology program; and learn from failures.7
technological and contextual similarities with DTx, can be used as a

proxy to learn from on DTx implementation:
• Assess how each of the technology characteristics, such as its

usability, adaptability to local contexts, and compatibility
with existing systems, may affect the effectiveness of the
implementation.
• Assess how the technology standards for interoperability, data
security, and data privacy can improve acceptability and
implementation (discussed in detail in chapter 6).
• Ensure there is sufficient financial and legislative support to
assist implementation.
• Identify key stakeholders and knowledgeable champions with a
track record of supporting and successfully advancing promis-
ing new health technologies, and include them early on in the
design, planning, and implementation process.

• Provide reliable educational and training materials about the
technology’s features, benefits, and workflow integration
scenarios aimed not only at primary users (i.e., physicians and
patients) but at all individuals and entities involved in paying
for and using the technology (i.e., employers, health plans, etc.).
• Support an overall cultural shift at the clinic/point-of-care that
promotes the benefits of technology and innovation.
• Take a proactive, systematic, and global view of the implemen-
tation process. It must be considered at appropriate stages in
the DTx development continuum given the multitude of
segments involved in its successful execution. A well-planned
implementation process ensures the state of readiness of end
users and their organizations, as well as provides critical user
feedback that can lead to dynamic procedural adjustments that
can catalyze or potentially preclude adoption. For example, new
roles supporting technology implementation may be required at
the clinic (as well as within the DTx organization), with multi-
disciplinary teams combining digital and clinical expertise.
Conclusion
Testing, implementing, and ultimately scaling up DTx are not easy
tasks. The implementation of DTx solutions does not conclude with a
going-live announcement or with a press release, as we see too often.
It should be considered a comprehensive process, not an event.
Following an IS approach and taking the implementation process
through a sequence of activities (such as involving health care system
managers early on) and appropriate support mechanisms w ill facili-
tate the adoption of interventions and empower the solution to work
at its full potential in a specific context. DTx solutions, like any other
innovation that aspires to be sustainably adopted over time, w ill need
to adapt to their unique local environments dynamically. As a result,
the DTx implementation strategy must be monitored and modulated,
and the underlying technology evaluated, on an ongoing basis to

ensure that intended goals are met, benefits are realized, and barriers
to effective use are identified and addressed.
Notes
1. NUHS is a group of Singapore health care institutions composed of several
hospitals, national specialty centers, and polyclinics.
2. Mathews and Yadron, “Health Insurer Anthem Hit by Hackers.”
3. Plsek, “Redesigning Health Care with Insights from the Science of Complex
Adaptive Systems.”
4. Implementation outcomes such as acceptability, adoption, costs, or penetra-
tion are indicators of implementation success but not of treatment effectiveness.
They should therefore only be considered as “intermediate outcomes,” which serve
as necessary preconditions for attaining subsequent desired changes in patient’s
clinical outcomes.
5. Ross et al., “Factors that Influence the Implementation of e-Health.”
cal F
7. Greenhalgh et al., “Beyond Adoption.”

• Chapter 10
Supporting Adoption of DTx
This chapter was coauthored by Dr. Robyn Mildon, Founding Executive

Director of the Centre for Evidence and Implementation with offices in
Australia, Singapore, and the United Kingdom.
The reality is that many of the envisioned solutions in health care are
not widely adopted or scalable (enough) to put into practice, and any
proposed changes to current practices—like a new technology—has
multiple barriers to overcome. W ill things be different for DTx? W
ill
this increasingly prevalent approach smoothly implement itself? Given
its intended role in transforming the delivery and administration of
medical care, this is unlikely. DTx organizations need to be proactive
in their implementation efforts and take learnings from other tech-
nologies that are being implemented and adopted (or not) in health
care. In this chapter, we provide two perspectives, those of health pro-
fessionals and t hose of patients, for how DTx can achieve buy-in. The
two often go hand in hand, in a virtuous cycle. We share here what
we believe are the key considerations and factors—often beyond the
technology itself—to enter that cycle to catalyze and sustain DTx
adoption.
173

Achieving Clinical Adoption of DTx
Acceptance by health care professionals may be the single most impor
tant determinant of whether a new health technology or serv ice suc-
ceeds.1 Understanding the reasons for potential resistance to using and
prescribing DTx and finding ways in which DTx organizations can sup-
port DTx acceptance and adoption among clinicians is, therefore,
imperative.
One of the physicians interviewed by our team noted, “Doctors are
early adopters of technology that works and that they understand.”
Physicians recognize the value of new drugs and medical device inno-
vations, such as small pacemakers. With increased exposure to digi-
tal health platforms, doctors are also becoming more accepting of
platforms into the patient care repertoire. According to a survey from
the American Medical Association, the use and appreciation of infor-
mation technology and digital health tools by physicians in general has
seen a substantial increase between 2016 and 2019. This includes the
category of remote monitoring and management for improved care
solutions, to which DTx belongs.2
Having said that, health care startups often fail to understand
physicians’ perspectives, which gives rise to some of the unforeseen
barriers to adopting DTx that the DTx designers discover down
the road. For example, with the technology evolving so rapidly and
algorithms-driven decision-making processes requiring more clarity,
physicians may need substantial support in sufficiently assessing
and, ultimately, trusting the technology. Additionally, digital health
is often associated with increased data management demands that
add more administrative burden to the clinicians’ daily practice, po-
tentially challenging or even discouraging adoption. Following is an
overview of the key issues that should be anticipated and addressed
proactively to engage and prepare physicians to embrace technology-
enabled health care.

Identifying High-Priority Needs
Empowering physician buy-in for an emerging technology starts with
demonstrated abilities to solve established challenges in their daily
practice. For example, for British primary care physicians, one of the
most common issues their patients present with, and for which there
is currently no agreed-upon treatment, is insomnia.
Insomnia is common and debilitating, yet despite 10% prevalence there is no gold
standard treatment in the UK that is readily available for prescription or referral
in the NHS [National Health Service]. There is a massive treatment gap that
physicians want to solve. At Big Health, we provided physicians with Sleepio, a
DTx for insomnia, backed by 12 randomized controlled trials—the gold
standard when it comes to scientific evidence. We also partnered with organ
izations to provide physicians with sleep and DTx education—how to assess for
insomnia, which patients are indicated for our DTx solution, how to support
patients, and when to follow up. The DTx has to be embedded into existing
clinical pathways, rather than being introduced as an “app.” —Dr. Charlotte
Lee, UK Director, Big Health, a DTx company that develops solutions for
mental health.3
DTx implementation requires early interest from physicians. As you

might expect, if your DTx solution offers to address physicians’ top
clinical priorities, you’ll have a better chance of securing their time for
a product demonstration and education sessions—key entry points for
subsequent pathways toward integration into a clinical workflow.
T hese sessions, also referred to as physician training and education,
provide opportunities to continuously communicate the results of vari
ous studies and clinical t rials regarding the safety and effectiveness
of solutions that are already mature. It’s the first step to gaining phy-
sician trust and avoiding physician bias.
It is also strongly advised to involve physicians as early as possible
throughout the DTx development and implementation stages. This
level of proactive engagement helps to ensure that DTx platforms
address their needs while also helping them to gain a foundational
Supporting Adoption of DTx 175

understanding of the technology and how to use it effectively. T hese
engagements w ill increase physicians’ overall confidence in DTx and
drive ownership, ultimately empowering physician adoption.
Anticipating Workflow Changes

Recent studies looking at how physicians divide their time daily found
that for e very hour dedicated to providing direct clinical care to pa-
tients, they spend up to three additional hours on administrative
work.4 How DTx technologies are incorporated into established clini-
cal workflows is therefore critical to ensuring that they do not create
additional burden for physicians. This is important as high patient
loads result in physicians spending a substantial portion of their time
looking at their screens and away from their patients, which can sub-
sequently impact the patient-physician relationship.
And just as we are now discussing the changes that a DTx may intro-
duce to existing workflows and communications, we also need to eval-
uate its compatibility with future workflows, future upgrades to the
infrastructure that would support it, and future needs of the health
care staff that would use and recommend it to their patients. DTx may
disrupt existing patterns of team interaction in ways that can prove to
be more complex than initially anticipated.5 The adoption of DTx often
involves multiple stakeholders, numerous decision-making processes,
potentially novel routes for communication between care teams or be-
tween care teams and patients and caregivers, as well as a series of per-
sonal and organizational value judgements. From past experiences with
introducing new technology into a clinical setting, we have observed
that clinical and administrative staff are vital when it comes to inte-
grating innovations into everyday clinic use. Consequently, patients are
more likely to take advantage of, and benefit from, such interventions if
clinical staff have been extensively involved with vetting and facilitat-
ing the onboarding process and if the intervention has been introduced
by staff members who recognized its capacity for markedly improved
patient outcomes over traditional approaches.6

Addressing DTx-Driven Changes in the
Patient-Physician Relationship
In the 1980s and 1990s, artificial intelligence (AI) was perceived as a
not-so-subtle threat represented by the Terminator and Matrix movie
series. But in 2020 and beyond, instead of taking the form of a men-
acing robot or omnipresent authoritarian, AI is more likely to take on
the form of recommended purchases on Amazon or targeted Facebook
ads. Concerns about technology’s impact on our everyday lives have
evolved too, expanding from the influences it can wield in our digital
and online environments to affecting how we engage in person-to-
person interactions, especially in health care. And in the realm of
health care, these fears can sometimes morph into worry that digitiz-
ing health care—whether by bringing in surgical robots or DTx—w ill
replace both physicians and the human connection they forge with
their patients. As a result of the workflow changes outlined in the pre-
vious section, it is necessary to consider how DTx may ultimately af-
fect the patient-physician interaction in some way.
Patients’ motivations for visiting a doctor usually encompass much
more than a desire to receive a medical report about their health state
and recommendations on how to continue to care for themselves. The
compassion, empathy, and reassurance they expect, and often receive,
from their physicians have been shown to have a substantial impact
on both the individual’s health outcomes and the health care system’s
own outcomes (including economic).7 T hese benefits would be too pre-
cious to lose if a digitally powered health care system was perceived
to somehow constrain such personal patient-doctor interactions. DTx
solutions should be carefully incorporated into existing health care
modalities so as not to disrupt any positive patient-doctor interactions
that have already been established but instead aim to support and po-
tentially even improve their quality (e.g., through increased accessi-
bility), where possible.
In an ideal DTx-empowered clinical practice scenario, such technolo-
gies would serve to automatically generate information about disease

evolution, assessment, and treatment plans, thereby freeing up time to
discuss what really matters to a patient—their life and how the disease
impacts it. By shifting most of the informational and educational work-
load from the physician and empowering the patient, DTx platforms
can facilitate the provision of care beyond basic medical needs, and as
such have the potential to improve human connections.
Addressing the Cost Issue

As mentioned, insomnia care may be a high-priority need for physi-
cians, yet during our interviews with health care professionals, one
physician, referring to the financing of DTx implementation and us-
age, promptly asked us, “Who w ill bear the cost?” It is a critical ques-
tion for any DTx solutions, and the solution starts with demonstrat-
ing the intervention’s cost-effectiveness. For example, a recent study
looked at the cost-effectiveness of digitally administered cognitive be-
havioral treatment (CBT) for insomnia (Sleepio).8 The study looked at
both direct and indirect costs— including treatment expenses,
insomnia-related health care expenses, and the costs of reduced pro-
ductivity at the workplace. The quality-adjusted life years (QALYs),
which is an established method of determining the value of health care
treatments, was designated by this study as US$50,000. This work also
compared diff erent forms of intervention against digitally adminis-
tered CBT (including pharmacologic intervention, group CBT, single
patient CBT, and no treatment). The study demonstrated that digital
CBT was the most cost-effective treatment, but it also had the highest
positive net monetary benefit (NMB) relative to no insomnia treat-
ment (i.e., $681.06 for each patient over a six-month period). More
specifically, the NMB term served as a measure of the value of a treat-
ment in a monetary context when there was an established threshold
for a willingness to pay. In other words, a positive NMB shows that
the cost of realizing the benefit is less than the maximum amount that
a decision-maker (e.g., insurance company) would be willing to reim-
burse to achieve this benefit.

Beyond the cost-effectiveness aspects, would the DTx be covered
under a specific benefit, whether pharmacy, medical, or even value-
based contracting? Would the patient’s insurance pay for it? Or would
it be covered by patients out of pocket? The question of cost, who bears
it, and the cost-effectiveness in the adoption of DTx is crucial. A sys-
tematic review of 221 studies looking at the f actors influencing suc-
cess and failure of e-health interventions found that cost was the most
frequently reported contributing factor to failure.9
Although new technologies in general and DTx in part icu lar are
commonly portrayed as long-term cost-saving opportunities, their im-
plementation and the learning curve they require for deployment of-
ten imply the opposite. Patients may still need to take their pills and
continue to see their physicians as usual while both sides get comfort-
able enough with the technology for it to replace the conventional care
pathway. This is due to the fact that some of the challenges that consis-
tently bedevil health care include staff shortages and a generalized
dearth of training, space, and equipment. Yet to fully integrate DTx
into new clinical workflows, health care providers need to “go digital”
by transitioning to appropriate infrastructures that drive digital ser
vices and developing other digital-centric capabilities.
In a nutshell, this is one of the few points on which all stakeholders
in the DTx ecosystem will agree: DTx must be cost-effective and aligned
with value-based medicine. DTx must therefore build a case for long-
term savings while also addressing the upfront expenses of the tech-
nology induction period and the ongoing costs.
Supporting Patient Adoption of DTx

Having the Right Engagement Strategy
here is growing literature on how engaged patients are more likely
T
to be proactive and empowered in their care, understand their condition,
and ultimately adopt new healthy behaviors—a profoundly important
and elusive outcome. The key finding from t hese studies is that higher
incidence of patient engagement is associated with improved health

outcomes.10 It also positively impacts the patients’ and clinicians’ ex-
periences and delivery of care. So what does an effective patient en-
gagement strategy look like in the digital age?
Today, the patient-physician interaction is frequently the primary
determinant of the patient experience. The personalization of this ex-
perience is often limited to adjustments to the treatment plan (e.g.,
changes to regimen, dose titration, etc.). A DTx, on the other hand, can
provide a broad spectrum of new digital channels and, as a result, new
digital delivery mechanisms to drive engagement with both providers
and patients. For example, providers can potentially tap into digitally
enabled channels to sustainably communicate and engage with pa-
tients to disseminate content or track symptoms. DTx developers can
also collect real-time patient information (e.g., self-reported outcomes,
patient feedback systems, emotion tracking, etc.) to engage patients
proactively and/or in a more timely and personalized manner, lever-
aging a wide range of data and information.
These new digital touchpoints and interactions across channels pro-
vide new engagement strategies. Personalized nudges, online support
groups, connected caregivers and families, bite-sized e-learning, digi-
tally shared decision-making interventions, and potential instant grat-
ification through gamification elements are just a few emerging pa-
tient engagement strategies, each with a diff erent level of maturity and
evidence. T hese resources serve as potential catalysts toward achiev-
ing truly personalized patient treatment as well as engagement expe-
riences. This can, in turn, empower patients, drive sustained use of the
DTx, and ultimately achieve better treatment outcomes compared to
conventional strategies. This is echoed by recent insights shared in our
contributor interviews following a COVID-19 use case. “We used to
send daily reminders to COVID-19 patients for symptoms reporting,
but we realized after some time that most patients had formed habits
and we r eally only needed to send them reminders when they forgot
to report their readings,” said Dorothea Koh, founder and CEO of
BotMD, which powers Singapore’s National University Health System
virtual wards program to care for COVID-19 patients remotely.11

One relevant approach to patient engagement for DTx is an omni-
channel approach that fully leverages these new touchpoints. It is de-
fined as a multichannel patient engagement strategy that w ill collect
and aggregate all clinical and nonclinical data and information gener-
ated by t hese new touchpoints and interactions. This integration of
systems, channels, and information creates a single view of the patient
and for the patient. It provides unique insights into their treatments,
also revealing a full picture of a patient’s health profile beyond only
clinical aspects. It also offers the possibility to take into consideration
the range of personal, social, economic, and environmental factors in-
fluencing a patient’s health. For example, emotions have been shown
to alter judgment and decision-making.12 So how can DTx account for
emotional ups and downs that patients experience to personalize the
engagement experience? Since there i sn’t a one-size-fits-all patient en-
gagement strategy, and offering patients individualized content and
intervention is essential for building stronger engagement, and prob
ably better outcomes, an omnichannel approach may serve as a suit-
able solution.
The omnichannel approach has been mastered by retail and e-
commerce players, offering a seamless shopping experience both on-
line and in-store. Integration across several touchpoints (i.e., physical
stores, websites, product pages, apps, mobile ads, and other means) is
critical for generating data, traffic, and, eventually, sales. For DTx,
there are two main goals, closely interconnected. The first goal is to
create a seamless and personalized experience that meets patients
where they are—the possibility to address issues that m atter to them
and respond to their preferences, needs, and values. The second goal
is to empower patients in their own care, increase adherence to a
therapy, and encourage behavior change, with the ultimate goal of
improving outcomes.
We should emphasize that incorporating patient engagement strat-
egies should not be considered only when patients are first prescribed
a DTx. In fact, these strategies should be addressed from the very be-
ginning of the solution’s design phase. It’s a multistage process from

pre-to post-intervention (i.e., activation, sustained use, and weaning
off the DTx when the treatment is finished). A robust and effective DTx
patient engagement strategy should also be developed with a user-
centric approach. For example, the Patient Centered Outcomes Re-
search Institute created a “patient co-investigator” role for their stud-
ies. Patients are more than study participants. In fact, they are vital
stakeholders empowering the research studies that could include the
design of a new digital health intervention that could impact their own
health journeys.13 As such, effective patient engagement approaches
can help DTx developers comprehend f actors and design elements to
drive sustained use of the solution and ultimately support positive user
experiences and outcomes.
Strengthening Relationships
In a health care context, DTx solutions should strengthen
relationships—the relationship between a patient and physician and
the relationship between the patients and their own health. T hese rela-
tionships play a vital role in driving and sustaining DTx adoption, and
reinforcing these relationships involves understanding the emotional
aspects that underlie the decisions patients make in consultation with
their doctors and on their own. For example, patients signing up for a
digital health program through a physician referral, an employee assis-
tance program, and an app store may have different expectations and
motivations. As Dr. Lee shared with us, “Patients first need to acknowl-
edge that they have a concern or a problem before deciding that they
need to see a physician. When this is acted upon, their motivation to
accept the recommended solution by their physician—be it a drug, an
app, or a referral to a different service—is high. The two elements,
trust and authority, are critical in this patient-physician relationship.
The challenge for DTx, such as Big Health, is how to take this motiva-
tion and transform it into something scalable and accessible.”
Trust is central to DTx uptake, as can be observed when would-be
users hesitate to embrace a technology if they feel it is “faceless” or

lacking a “human touch.” Fortunately, this trust can be fostered by a
third party; evidence shows that seeing a trusted person—a friend,
family member, or a highly visible public figure—already using the DTx
would often lead to interest in and engagement with the program. For
this reason, Big Health decided to involve “link workers”—nonclinical
social care practitioners embedded in primary care practices—to en-
gage with and support patients who have been prescribed their apps.
According to Dr. Lee, the retention rate for Sleepio, the company’s DTx
for poor sleep, r ose substantially in Scotland with that model.
Trust in DTx solutions can also be engendered through connecting
with other users of the same technology. Some companies are offering
patients the opportunity to be part of a peer community as a built-in
functionality of their platform and user interface. This approach
has been found to be an effective method of improving engagement
and achieving higher adherence rates compared with unsupported
interventions.14
As a case in point, in one study looking at the effectiveness of vir-
tual reality–based rehabilitative stroke therapy, researchers found that
stroke patients training in the multiuser version of the system, which
provided the ability to connect with o thers online, showed greater
adoption compared to the same patients training in the single-user
mode. Specifically, patients spent up to 22% more time when using the
multiuser version.15 In another study, 30% of tobacco users who en-
rolled in a digital program that worked with a health coach managed
to quit for good, compared to the 20% who did so a fter engaging only
with digital tools.16
Many DTx today, in our opinion, lack an essential hook as part of
their offering—t he ability for users to connect with someone like
them, someone who could empathize with their health care journey
because they have the same health condition or lived experience with
a disease. PatientsLikeMe, an online community where patients ex-
change thoughts and advice for managing a variety of diseases with
others, embodies the concept well. However, this community differs
from a DTx. Peer support, including online peer support communities,

is gaining popularity as a means of supplementing the self-management
of chronic health conditions. It can be a valuable source of informa-
tion and a safe haven for patients, frequently extending beyond med-
ical care. As proof of their effectiveness emerges, expect DTx to em-
brace such features.
Celebrating Success with Short-Term Wins

Remember Fabien, his daily struggles to manage his type 1 diabetes,
and his rollercoaster of emotions? His physician advised him that his
main objective was to manage his blood sugar levels—not the most
exciting or engaging task from Fabien’s perspective. Frequent remind-
ers from the physician and from his family about the risk of suffering
kidney failure or a foot amputation if he fails to manage his condition
well was not helping his morale, e ither. To stay motivated, he needed
concrete, trackable, and achievable short-and long-term objectives. As
he noted during our interview, “What if I want to climb Mount Everest?
What if this is my objective and not managing diabetes or getting my
blood sugar u nder control? Tell me what I need to do, the steps required
so that I can ultimately be successful at climbing Mount Everest.”
Indeed, Fabien’s definition of success, which is to climb Mount Ever-
est, will almost certainly not coincide with his physician’s definition of
success, which is to improve his health outcomes, clinically speaking.
Success from the perspective of a physician caring for a diabetes patient
is primarily based on lowering the patient’s glycemic variability, while
from the perspective of the patient, success is based on managing their
condition with as little effort as possible so that they can accomplish
other personal or professional goals that are meaningful to them while
having a minimal effect on their daily routines (figure 10.1).
As definitions of successful disease management differ depending
on who we ask, we invite DTx designers to consider both patient and
physician perspectives on how a DTx solution could benefit each party
and help them achieve their respective goals and objectives. In addi-
tion to taking the long view, DTx designers also need to ensure their

Improved Glycemic Lifestyle Therapy
Outcomes variability flexibility effort
Physician’s priority Patient’s priorities
Figure 10.1. Improved outcomes for a diabetes patient is a combination of both

physician priorities and patient priorities. Source: From the authors
technology is delivering enough short-term wins—symbolically cele-

brating hitting predetermined milestones along the way—in order to
keep users involved. T
hese acknowledgments could be designed to take
place right when the DTx is introduced to the patient for the first time,
within the first few weeks of them using the therapy, or periodically
during the course of treatment. The idea is to provide psychological
reinforcement for any positive changes in behavior, symptoms, or clin-
ical outcomes that the DTx is able to accurately capture and track
longitudinally.
Nudge Me Right!
uman nature does not always produce the rational decision-makers
H
we may hope to be. When it comes to managing even our own health,
we don’t always have the willpower to stay the course on a treatment
regimen, exercise, or diet plan. Rigorous cost-benefit analyses and
trade-offs between immediate gratification and future health out-
comes often go out the window when we face the option of d oing some-
thing that puts a physical or psychological burden on us, as one might
find with trying to adhere to a workout plan or a demanding daily ther-
apy. Luckily, t hese natural biases in the decision-making process can be
turned into an advantage by encouraging health-enhancing behaviors
through the use of “nudges.”
The word nudge was popularized when the Cameron government in-
troduced the Behavioural Insights Team in the UK—also known as
the Nudge Unit—to investigate how desirable behavior could be en-
couraged. In a nutshell, nudging attempts to subtly lead people toward
making the right decision. It can be used to encourage behaviors that

are beneficial to one’s health compared to a common default behavior
that could be construed as a bad habit. For example, instead of out-
right banning junk-food advertising, nudging can change the order in
which options are presented, with healthier options listed first to sub-
consciously steer consumers’ attention to healthier choices.
The same principle can be applied in the context of making health-
related decisions. How can patients’ environments be tweaked so as
to help them make healthier choices? For example, changing a choice
architecture default by requiring people to opt out of organ donation
rather than opt in may lead to more organ donations and more lives
saved.17 In the same vein, Penn Medicine in 2014 switched its electronic
health record system’s default from brand-name drugs to their generic
equivalents. This immediately increased generic prescribing rates from
75% to 98%, leading to a $32 million savings to both patients and the
health system within two and a half years.18 Common behavior mech-
anisms used to nudge patients are listed in table 10.1.
Closer to DTx, Livongo—which offers chronic disease management
programs, including for diabetes by combining connected devices, data
sharing, and coaching—is a good example of a digital health technol-
ogy built on the principle of nudging. By the end of 2019, the company
had delivered 2 million “Health Nudges,” which they defined in their
preliminary prospectus with the US Securities and Exchange Com-
mission as “small, readily accessible interventions, which can include
hundreds of different behavior or lifestyle adjustments that can alter
clinical outcomes for p eople with a chronic condition, such as diet ad-
vice, medication information, or suggestions for increases in physi-
cal activity. People simply want to live their lives, not be constantly
reminded that they are living with a condition.”19
Livongo’s health nudges work by detecting a behaviorally modifiable
issue that could lead to worse health outcomes if left unaddressed
and sending out a suggestion for corrective action. For example, when
the system identifies a user that has not been checking their blood
glucose in the mornings, a nudge is sent to encourage them to do so
before breakfast in order to better understand overnight patterns.

Table 10.1. Behavior mechanisms to nudge patient
Changing Individuals prefer sticking to existing or standard behaviors rather than doing
defaults something different or involving an effortful choice.
Example: Opt-in vs. opt-out strategies for organ donation.
Hyperbolic Individuals prefer short-term rewards, do not like uncertainty and w ill try to reduce
discounting it whenever possible, and tend to undervalue less concrete f uture rewards.
Example: Rewarding an increase in steps, logging blood glucose, and light
exercise vs. an HbA1c reduction after three months.
Loss and risk Individuals are reluctant to take risks and accept potential losses, unless this
aversion can be compensated by potential important rewards. In other words, they do not
mind not having something, but they do mind losing something.
Example: Whenever attempting to change patient behaviors, health care
professionals need to be able to explain the benefits and gains to their patients,
and contrast it with the losses, so that t hese can be tangibly assessed.
Framing Decision-making is perceived as easier when a few options are available, but at
the same time, having too many options can be counterproductive.
Example: Choices should always be framed in ways that help patients and staff
understand what their preferences are.
Reciprocity Individuals strive to reciprocate commitments. They are more likely to change
their behaviors if they feel that they owe someone e
lse something.
Example: Studies that applied this principle to health care issues found that
organ donations could be increased significantly after a campaign containing
the message “If you needed an organ transplant would you have one?”
Social norms Individuals tend to base their behaviors on what they perceive others are doing,
and feedback comparing themselves and what they think they are expected to do in order to
(or peer conform to the norm. Feedback on how individual performance evolves over
pressure time can help while also improving behaviors.
influence) Example: Instead of saying “Please arrive on time,” state that “80% of patients in
this clinic arrive on time for their appointments.” Communicating the percentage
of patients who arrive on time for their appointment has been showed to
decrease no-shows by as much as 30%.
Source: Voyer, “ ‘Nudging’ Behaviours in Healthcare.”
hese interventions are provided through the member’s preferred

T
communication channel (e.g., a blood glucose meter, a mobile app, the
web, or email) and are based on an analysis of patient-generated infor-
mation collected over time, such as self-reported data and data gener-
ated from wearables and sensors. The analysis consists of scanning for
patterns that help to select the type of nudge that would be most effec-
tive in a part icu lar context and for a part icu lar individual. Livongo

noted that its health nudges lead patients to make smarter health
choices and drive behavior changes: “Over 40% of our members who
received a Health Nudge changed their behavior in response to a sug-
gestion that they complete a blood glucose check before breakfast.”20
Penn Medicine and Livongo’s use of nudges demonstrate how strat-
egies and tools borrowed from the field of psychology can be lever-
aged to influence patient behavior. However, nudging is just a tool that
is part of the solution to better disease management, but not the solu-
tion. A multifaceted approach is necessary to create an engaging tech-
nology that steers patients to make healthier decisions, and motivates
physicians to endorse and prescribe new DTx solutions.
Conclusion
As more data emerge on user engagement with DTx, it is critical for in-
novators to take these into account and respond adaptively by remov-
ing, adding, or improving upon features that lead to better motivation,
sustained use, and clinical outcomes. To achieve this, it is important to
involve physicians and patients as early as possible, where appropriate,
in the development and deployment of DTx. This cocreation approach
to explore existing problems and potential solutions holistically can
substantially improve the likelihood of a DTx platform being a dopted.
Put simply, people support what they have helped create.
Notes
2. Miliard, “AMA Sees Surge in Health IT Adoption, ‘Rise of the Digital-Native
Physician.’ ”
3. From an in interview with the authors, October 26, 2021.
4. Wenger et al., “Allocation of Internal Medicine Resident Time in a Swiss
Hospital”; Sinsky et al., “Allocation of Physician Time in Ambulatory Practice.”
6. Aref-Adib et al., “Factors Affecting Implementation of Digital Health Interven-
tions for People with Psychosis or Bipolar Disorder, and Their Family and Friends.”
7. Trzeciak and Mazzarelli, Compassionomics: The Revolutionary Scientific
Evidence that Caring Makes a Difference; Derksen, Bensing, and Lagro-Janssen,
“Effectiveness of Empathy in General Practice.”

8. Darden et al., “Cost-Effectiveness of Digital Cognitive Behavioral Therapy
(Sleepio) for Insomnia.”
9. Granja, Janssen, and Johansen, “Factors Determining the Success and
Failure of eHealth Interventions.”
10. Greene et al., “When Patient Activation Levels Change, Health Outcomes
and Costs Change, Too”; Krist et al., “Engaging Patients in Decision-Making and
Behavior Change to Promote Prevention”; Bombard et al., “Engaging Patients to
Improve Quality of Care.”
11. From an interview with the authors, May 19, 2022.
12. Lerner et al., “Emotion and Decision Making.”
13. Robbins, Tufte, and Hsu, “Learning to ‘Swim’ with the Experts.”
14. Aref-Adib et al., “Factors Affecting Implementation of Digital Health
Interventions for P eople with Psychosis or Bipolar Disorder, and Their Family and
Friends.”
15. Thielbar et al., “Home-Based Upper Extremity Stroke Therapy Using a
Multiuser Virtual Reality Environment.”
16. “VP Live,” Virgin Pulse.
17. Davidai, Gilovich, and Ross, “The Meaning of Default Options for Potential
Organ Donors.”
18. Levins, “Report from the First National ‘Nudge Units in Health Care’
Symposium.”
19. Livongo Health, “United States Securities and Exchange Commission Form
S-1 Registration Statement,” 3.
20. Livongo Health, “United States Securities and Exchange Commission Form
S-1 Registration Statement,” 131.

• PART IV
PATHS TO COMMERCIALIZATION
This part was coauthored with Christopher Hardesty. Hardesty is a

partner at Pureland Venture in Singapore who specializes in building
early-stage medical innovations for scale-up in Asia and beyond. He
has lived and worked in more than 50 countries to design public-
private health system financial schemes, including for the creation
of pathways for the adoption of medical innovations.
Pre-COVID-19, we knew a fundamentally new approach was required

to transform our health care systems and respond to the rise in
chronic and mental health diseases and their consequences. Regula-
tors, payers, and prescribers were starting to learn about DTx as a
valid therapy alternative or complement for the treatment of a variety
of illnesses.
With the recent adoption of digital health and models for deliver-
ing health care remotely, we have seen a variety of new commercial-
ization strategies being undertaken by early-stage and established
manufacturers of DTx. Since the COVID-19 emergence, the US Food
and Drug Administration (FDA) has exempted many regulatory
requirements for DTx treating psychiatric disorders, such as the
need to file a 510(k) premarket notice. Within a week, Akili Interac-
tive made its DTx for ADHD available to qualified families at no cost
and without the need for a doctor’s prescription. Simultaneously,
Corey McCann, CEO of Pear Therapeutics, stated that teleprescrib-
ing for its therapies had “gone through the roof.” The pandemic did,
in fact, generate new collaboration possibilities and care models
as individuals sought care in places other than hospitals, such as
191

patients’ homes and communities. In summary, the broader health
care ecosystem is appreciating more and more the potential of DTx.
Is a health care revolution underway?
In 2019, the global DTx market was valued at $2.69 billion.1 In 2020
alone, several DTx solutions achieved meaningful early scale, includ-
ing new rounds of funding with $1.2 billion across 52 deals in 2019,2
including mergers, partnerships, and high-profile product launches in
the United States, European Union, and Asia. According to Med-
Rhythms, the DTx industry went from having about 13 new products
a year between 2012 and 2015, to 44 a year between 2016 and 2019, to
80 in 2020 alone.3 There was also a significant uptake of DTx for
neuropsychiatric indications from 2012 to 2020 (figure Part IV-1).
The global DTx market is expected to reach $11.82 billion by 2027,
growing at a compound annual growth rate of 20.3% during the
2020–2027 period.4 DTx are certainly gaining traction, but what
would it take for DTx to achieve meaningful scale and play a pivotal
role in the transformation of our health systems?
Despite the notable successes of a few DTx technologies in terms of
conceptualization, regulatory clearance, and early adoption, all too
often achieving scalable commercial success remains a hard-to-reach
reality. Perhaps the best known example of the elusive nature of
success for DTx organizations is Proteus, the “smart pill” startup that
Overall average new DTx Average number of new DTx per category
per year of first mention for each period per year of first mention for each period
Neuropsychiatric
25
Other
20
80
15
Neurologic
44 10
Musculoskeletal/Pain
5 Cardiovascular/Metabolic
13 Respiratory
Dermatologic & Wound
0
2012-2015 2016-2019 2020 2012-2015 2016-2019 2020
Figure Part IV.1. Proliferation of DTx products. Source: McCarthy, “Looking Back &
Ahead at the Digital Therapeutics Industry”
192 Paths to Commercialization

was once considered a unicorn b ecause of its US$1.5 billion valuation
but then filed for bankruptcy in 2020.5 For nearly two decades,
Proteus had planned to revolutionize medication adherence with its
ingestible sensor—a tiny microchip the size of a grain of sand that
could be planted into any drug and send signals to a smartphone app
and a digital portal to inform if the pill was swallowed or not. But as it
sought to cross the chasm between being a new, little-known explor-
atory product and becoming a widely a dopted recognized solution,
the organization suffered from two common blind spots:
1. Lack of full channel understanding. In the case of Proteus, the

eagerness to quickly go to market led the company to prema-
turely strike a deal with the Japanese multinational company
Otsuka Pharmaceutical. The objective of the deal was to
become a leader in the emerging category of “digital medicines”
by pairing up with Otsuka’s popular antipsychotic drug Abilify
(aripiprazole), approved for treating various neurological
conditions in adults. T here w ere many challenges with this
drug–device pairing, such as the fact that people with schizo
phrenia, for whom Abilify is frequently prescribed, tend to fear
being controlled or monitored, which in a way is what Proteus’s
DTx was designed to do. Even if the partnership made techno-
logical sense, the company lacked a comprehensive understand-
ing of motivations and potential objections of the sensor-
equipped drug’s primary users, influencers, and purchasers,
including the various partners who would help distribute and
commercialize the solution (the channel partners).6
2. Value exchange that is not fit for the purpose. Any DTx innovator
seeking to commercialize their invention must eventually
grapple with the pricing question. Exacerbating the lack of
channel understandings for Proteus—referring to its distribu-
tion system—was the fact that Abilify cost roughly 80 times
more than cheaper drug alternatives. Therefore, the task for
Proteus to demonstrate equivalent if not better health outcomes
Paths to Commercialization 193

and economic returns became essentially insurmountable.
As the FDA recognized in the wake of the drug’s commercial
failure, “Regulatory approval for this first-ever digital drug
was based on weak evidence, with no evidence of better
adherence with the digital version of aripiprazole compared
with the non-digital version. Both the scientific literature
and news reports conveyed an unsupported impression of
benefit.”7
Being on the front line of health care innovation is never easy, as

the Proteus case demonstrates. Rather than being disheartened, let
us seek to learn from such prior experiences in order to improve the
scalable success rate of what is an unavoidable positive trend t oward
DTx innovation in the health care industry. In chapter 11, we focus
on channel partnerships as one of the most effective paths to
commercialization and discuss channel partnership strategies by
types, mechanisms for engagement, and best practices therein. The
goal of channel partnerships is to optimize a DTx product’s distribu-
tion and commercialization strategy, and we provide suggestions on
how to maintain momentum and ramp up your commercialization
journey as you hit adoption milestones along the way. In chapter 12,
we focus on the payers. We explore for the concept of value, espe-
cially the economic value of DTx, and how it might be demonstrated
to private and public payers. As value lies in the eye of the beholder
and is market-specific, it is critical to understand how such value is
defined from various payers’ perspectives, and therefore the evi-
dence that w ill be required to convince potential payers of the
necessity of a DTx solution.
Notes
1. Precedence Research, “Digital Therapeutics Market Size to Hit US$ 11.82
Bn by 2027.”
2. Bridge Point Capital, “Advancement of Digital Therapeutics Leads to a Huge
Market with Enormous Potential.”
3. McCarthy, “Looking Back & Ahead at the Digital Therapeutics Industry.”

4. Precedence Research, “Digital Therapeutics Market Size to Hit US$ 11.82
Bn by 2027.”
5. Farr, “Proteus Digital Health, Once Valued at $1.5 Billion, Files for Chapter 11
Bankruptcy.”
6. Muoio, “Proteus Parts Ways with Otsuka as It Pivots toward Oncology,
Infectious Disease Treatment Adherence.”
7. Cosgrove et al., “Digital Aripiprazole or Digital Evergreening?”
Paths to Commercialization 195

• Chapter 11
Channel Partnerships
The Ultimate Commercialization Strategy
fter considering how DTx adoption might be predicted and assisted

A
in the previous chapter, it is time to consider the significance of DTx
partnerships in the commercialization of the solution. The go-to-
market models may be quite particular to your therapeutic area and
intervention, but they may also differ from market to market.
To date, three business models have emerged as dominant in the
digital health and digital wellness space: business-to-consumer (B2C),
business-to-business (B2B), and business-to-business-to-consumer
(B2B2C). B2C is a model of choice for technologies that do not require
regulatory clearance, are not prescription-based, and can be purchased
“over the digital counter” of an app store by the end users. B2B and
B2B2C are models of choice for technologies that undergo higher level
of scrutiny and that have a differential approach to reimbursement.
Those models are nonexclusive and both require a structured approach
to planning, organizing, and executing the business strategy.
We have seen DTx innovators opting for not one channel but a mul-
tichannel distribution approach and therefore pursuing more than
one business model, e ither concurrently or sequentially. This approach
allows one to reach a range of users and diversify revenue streams,
197

which is of particular importance for companies that have just brought
a product to market. For example:
• AEvice Health, the previously mentioned Singapore-based

remote-monitoring company operating in the field of respira-
tory illness, is pursuing the B2C route so as to gain experience
and build a foundational dataset while planning for a regulated
medical solution that w ill use distribution partners for scale-up
support (B2B2C).
• DarioHealth, a DTx solution for chronic conditions, a fter
gaining momentum in B2C models and showing the required
clinical results at scale, transitioned to a B2B2C strategy,
targeting employers, health plans (i.e., any program that helps
patients pay for medical expenses, whether through privately
purchased insurance, social insurance, or a social welfare
program funded by the government), and providers.
• Glooko, a US-based software-as-a-service solution for diabetes
management, launched both an FDA-cleared insulin dosing
system paired with a mobile app for consumers (B2C) and a
“kiosk”-style offering for in-clinic health care professionals
that provide care for diabetes patients (B2B2C).
As you can see, the B2C model is a common way of introducing the
company and the product to the market. However, while some level
of direct selling may still be feasible at scale, true scalability across
geographies, customer bases, and payment archetypes typically re-
quires strategic partnerships with third parties. The good news is
that identifying such entities may be easier now than it was before
the pandemic. In the post-COVID-19 era, one often gets the impres-
sion that “every” business is pivoting into the health care industry,
and whether that is justified or not, the positive artifact is that the
universe of potential channel partners for DTx innovation is grow-
ing wider (table 11.1).
Each of t hese potential partners brings diff erent assets, capabilities,
and expertise. While some may be more suited to assist a DTx organ

T able 11.1. Potential channel partners for large-scale DTx distribution and
commercialization
Near-term mature • Life sciences organizations: mainly pharmaceutical and medical

commercial partners device companies and their distributors
• Employers: typically, medium-to large-size private companies
providing health insurance to their employees
• Insurers: mostly private health and life insurers
Medium-term emerging • TeleX organizations: including telehealth, telemedicine, and
commercial partners telecare organizations
Longer-term emerging • Telcos: large telecommunications firms investing heavily in
commercial partners health care
(subject to country • Big Tech: large technological companies seeking expansion to
specificities and readiness) the area of health care
ization’s journey toward broad reimbursement coverage, others pre

sent an alternative pathway to business model optimization and scal-
ability. Still, regardless of their orientation, it is necessary to keep in
mind that not all partnerships lead to the same revenue generation,
as some revenue models are still being determined, and therefore carry
an inherently higher risk.
Ultimately, a DTx organization must do its due diligence to deter-
mine whether partnering with a particular channel partner is a good
strategic fit relative to its planned commercialization journey. To help
frame how each partnership category should be evaluated, we cover
their distinguishing characteristics, advantages, and disadvantages in
greater detail next.
Life Sciences Organizations: The Natural Choice

Pharmaceutical Companies
The pharmaceutical industry is being disrupted. Pricing pressures,
patent expiries, desire for greater patient and clinician engagement,
and new forms of competition are forcing the industry to reevaluate
its traditional commercialization models. The blockbuster model, in
which hopes hinged on the extraordinary commercial success of a few
Channel Partnerships 199

Figure 11.1. Examples of pharma–DTx partnerships for DTx development (top row),
DTx commercialization (middle row), and DTx development plus commercialization
(bottom row). Source: From the authors
billion-dollar-selling drugs, is slowly being challenged by real-world,

data-driven, digitally supported therapeutic approaches like those
embodied by DTx startups.
Having seen the writing on the wall, pharma companies are increas-
ingly positioning themselves as a natural partner for DTx organ
izations throughout the clinical development, regulatory, and reim-
bursement pathways (figure 11.1). Granted, the go-to-market journey
of a DTx product is very similar to that of a drug asset in terms of de-
velopment, approval, commercialization, prescribing, and payment,
so this type of partnership is the closest there is to a perfect match.
Pharma companies naturally see DTx partnerships as a win-w in
strategy, particularly when there is a close match between a drug and

a software platform (a combination known in the industry vernacu-
lar as “drug-plus,” whose aim is to improve clinical value by enhanc-
ing the drug’s efficacy). One reason for that is such partnerships enable
pharma organizations to differentiate themselves from competitors.
In that frame of thinking, despite the eventual dissolution of the
Proteus-Otsuka “marriage,” the intentions that led to it were right—
Abilify, which accounts for one quarter of Otsuka’s total sales, lost its
patent in 2015 and required an upgraded, differentiated approach to
retain customers.1
Pharma industry players also need differentiation when it comes to
engaging clinicians and key opinion leaders. Drug-plus programs or DTx
that are positioned as a value-added service to existing therapies can be
an interesting avenue to differentiation, on the condition that they can
demonstrate superior value compared to conventional therapy alone.
When looked at this way, DTx represents for pharma mainly some form
of a companion platform rather than a new business stream. Hence, this
type of pharma–DTx partnerships can take the form of copromotion or
distribution arrangements instead of full-scale acquisitions. Much like
in dating, the goal for both partners is to find an optimal strategic fit
rather than base the u nion solely around financials.
Beyond differentiation, an ability to offer drug-plus programs fits
into an evolving landscape of a health care solution offering. As the
industry expands its focus from the clinical effectiveness of products,
considered in isolation, to how t hose products fit into the broader chain
of value-based care, engagement with patients and physicians plays a
critical part. When done right, it can lead to better treatment adher-
ence and, therefore, better health outcomes, effectively increasing a
therapy’s effectiveness. And engagement is becoming increasingly vir-
tualized. Companion platforms provide an opportunity to increase
patient engagement (remember our e arlier discussion of nudges and
behavioral change?), which in turn can have a direct positive revenue
impact, since improved adherence tends to lead not only to better clin-
ical outcomes, but also to higher sales. Through carefully structured
data-sharing agreements, pharma companies may also gain the ability

to obtain new data and insights about the clinical impact of their
therapies across the care pathway; this is made possible by the data
collection mechanisms embedded by design in DTx technologies. T hese
real-word data, mediated through patient engagement and convertible
into real-world evidence, have clinical as well as commercial value.
It is for all of these reasons that pharma players are often keen to
acquire or partner with DTx startups from early development through
to clinical validation and launch. T here is a caveat, however. The rev-
enue models for both—meaning the economic benefits for both play-
ers, DTx and the pharma—are yet to be demonstrated.
Medical Devices Companies

Like traditional pharma organizations, medical device companies
(medtech) are similarly pressured into innovating their business mod-
els and their offerings. The Asia Pacific Medical Technology Associa-
tion (APACMed), which brands itself as “the voice of MedTech in Asia,”
has seen several of its corporate members make significant invest-
ments in the DTx area. The association points to ResMed’s acquisi-
tion of Propeller Health, which aims to pair ResMed’s cloud-connectable
medical devices for the treatment of sleep apnea and chronic obstruc-
tive pulmonary disease (COPD) with Propeller’s DTx solutions for
people living with COPD and asthma, as an example. Another collab-
oration, between Abbott and Sanofi, aims to simplify how people with
diabetes manage their condition with smart pens, insulin titration
apps, and cloud software.2
As medical device organizations are showing growing interest in
partnering with DTx companies, in a parallel move a number of DTx
and aspiring DTx companies have also joined the ranks of APACMed.
Naluri and Wellthy Therapeutics, which provide clinically validated be-
havioral interventions for chronic disease management, are two of
those early movers. While medtech and DTx have technological com-
monalities, both industries confront comparable obstacles in commer-
cializing their “Software as a Medical Device” solutions.

Aside from the shift of medtech business models, medical devices
are simply at the forefront of the digital revolution, with DTx being seen
as attractive opportunities for more “connected” devices and toward
better clinical outcomes. Many medtech companies position them-
selves as an attractive partner to DTx organizations that needs to be
won over into the partnership. DTx entrepreneurs would be wise to
develop their value proposition around how their solution provides
value to the potential medtech partner. This is especially pertinent in
the area of care pathways in which connected medical devices can play
a role in patient monitoring and treatment adherence.
Delivering Value through Traditional Distribution Services

Finally, we must consider the life sciences organizations’ distribution
partners, including pharmacies, but also pharmaceutical distributors,
such as McKesson, Zuellig Pharma, DKSH, and Takeda just to name a
few. Squeezed between advanced, expensive therapies marketed by
pharmaceut ical companies and heightened pricing pressures by the
payer and provider community, distributors must thread the needle
of arriving at acceptable market prices while grappling with a complex,
evolving supply chain of new health care solutions. Seeking to create
value in such a dynamic marketplace, t hese distributors—traditionally
seen as mere “box-pushers”—have been reinventing themselves from
pure drug and medtech purveyors into integrated health care serv ices
groups. Some now offer full-suite health care serv ices companies that
include logistics, clinical t rials, marketing, sales, insurance, and even
digital and disease management solutions.
Zuellig Pharma, for example, is one of Asia’s largest distributors of
pharmaceutical products, serving over 350,000 medical facilities
across 13 markets. Zuellig’s transformation included a US$50 million
investment in a Singapore-based innovation center in 2017 to develop
data, digital, and disease management solutions that w ill assist doc-
tors in developing more effective treatment plans, patients in manag-
ing chronic conditions, and payers in managing health care costs. In

2019, they also acquired the health tech firm Klinify, which provides
a cloud-based digital medical assistant that assists over 800 Malaysian
doctors in managing their clinics. The company stated that their ob-
jective is to leverage digital channels to supplement their existing com-
panies and connect the health care ecosystem.3 W hether through its
own in-house development or through a variety of partnership deals,
Zuellig, in addition to the reach and scale strategy, is building relevant
and complementary capabilities to DTx solutions.
Final Thoughts on Life Sciences Organizations

Life science organizations have the infrastructure, market intelligence
and expertise, dedicated salesforce, established relationships with key
opinion leaders, payer contracting know-how, and evolving distribu-
tion systems that DTx ventures need to achieve scalability and wide-
spread adoption. As such, they are ideally positioned as a natural chan-
nel partner for digital health innovators. Yet, despite the numerous
benefits such partnerships may bring to DTx entrepreneurs, they also
harbor risks.
Too many such cases as the previously discussed Proteus–Otsuka
partnership, show how pharma–DTx collaborations may ultimately fail
to achieve expectations if they originate from overexcitement and are
not supported by a solid market research effort and business strategy.
Pharma companies are themselves constantly evolving u nder the
aforementioned pressures, so sometimes the existing salesforce incen-
tives and key performance indicators, such as t hose aligned to the
“blockbuster drug” mentality, are diverting leaders’ attention away
from championing the development of alternative approaches to sell-
ing a digital solution. Lack of firm commitment to resources, aggres-
sive timelines, l imited budget, and cross-marketing can also stifle po-
tentially promising collaborations with DTx stakeholders.
On the other hand, DTx organizations’ contributions to the pharma
industry’s financial performance is yet to be firmly demonstrated. The
reason is that DTx products are not yet considered a turnkey solution

that can be easily implemented into established business processes nor
easily monetized over the next decade. Instead, their successful inte-
gration and commercialization requires iterative development and
evidence generation.
Nevertheless, we remain optimistic about pharma, medtech play-
ers, and their associated distribution channels to serve as key com-
mercial partners for DTx innovators. Partnerships between DTx and
life sciences organizations must become entrenched and long-term,
not just a step to get to the next round of funding. For this reason,
Pharma: A Partnership Example

Akili Interactive Labs, a DTx company that provides a prescription-based digital
treatment for ADHD using a gamification approach, is an example of a company that
concurrently pursues B2C and B2B2C distribution channels, depending on the market.
In 2019, to optimize their expansion to Asia, Akili announced a strategic partner-
ship with Shionogi, a Japanese multinational pharmaceutical company specializing
in treatments for pain and central nervous system disorders, to develop and
commercialize DTx for the Japanese and Taiwanese markets.
Akili reiterated their desire to build a direct-to-consumer distribution model, with the
announcement of a new Global Access Platform—a research and development as well
as commercial platform—for all global product development activities. The platform
also serves as an independent prescription and patient support system for DTx. Akili
retained control of the core intellectual property platform that supports the commer-
cialization of the DTx, as well as maintained stewardship of the data collected, in line
with privacy standards. On the other hand, Shionogi is responsible for regulatory filings
and has exclusive rights to the clinical development, sales, and marketing of the first
two Akili DTx for ADHD and autism spectrum disorder in Japan and Taiwan.4
In short, the partnership sought to leverage each company’s area of expertise in
terms of building a new class of treatment for patients while ensuring a sound
commercial model. This partnership for Akili probably demonstrates the necessity to
have a multipronged approach, market by market, for the distribution and commer-
cialization strategy of DTx.

DTx innovators are advised to choose their commercial and distribu-
tion partners wisely, and to retain strong cultures of innovation.
Employers: The Need for Speed?

In the countries that comprise the Organisation for Economic Co-
operation and Development, approximately 65% of the working-age
labor force are employed.5 In the context of our discussion, this sug-
gests that employers are a critical conduit for addressing the majority
of working p eople’s health and therapy needs. Employers, particularly
in the United States and Asia, are increasingly expected to serve as
health care enablers, mainly in the form of sponsoring health insur-
ance, but also by providing prevention, wellness, and disease manage-
ment solutions. In addition to footing the bill for many direct costs
for health care and lost wages, respectively, employers and employees
are collectively confronted with challenges associated with medical
leave that can range from temporary to prolonged disability. Depres-
sion, for example, is estimated to cause 200 million lost workdays each
year in the United States at a cost to employers of up to $44 billion.6
Employers benefit from a healthy, motivated workforce, both di-
rectly (e.g., lower insurance premiums) and indirectly (e.g., higher
productivity). To further cultivate this work environment, employers
started offering more general wellness platforms and lifestyle solu-
tions to drive employee engagement and increase their general health
literacy t oward health screenings and preventive health. For example,
while the impact of COVID-19 on patient health from an infectious dis-
ease standpoint was a huge challenge in itself, the pandemic’s role as a
driver of mental health challenges may not be anywhere close to being
fully realized. With the emergence of the pandemic, DTx solutions for
mental health have gained substantial visibility with the community
at large, but also with employers. T hese solutions are now available in
many industries to keep employees engaged, safe, and healthy.
In our discussion with the global consulting company Mercer, they
also confirmed a noticeable uptick in employers looking for new health-

focused solutions, including DTx, that may lead to greater employee
productivity and well-being, while also attempting to rein in rising
health care costs. As it is important for employers to prevent the on-
set of debilitating conditions for their employees, DTx winners ulti-
mately connect health care data to claims experiences that show im-
proved outcomes with lower costs.
Such DTx–employer partnerships are becoming increasingly com-
mon. Livongo, the aforementioned US-based chronic disease manage-
ment platform, has deployed a strategy whereby, via an intentional part-
nership network, it has formed bonds with employers and labor unions
and was ultimately integrated into associated health plans offered to
employees. At last report, Livongo had approximately 800 B2B partners
and was serving 220,000 members. Even the US government’s Em-
ployee Health Association, an insurance provider for federal employees,
is a customer, accounting for 45,000 of Livongo’s individual users. Not
surprisingly, the company is aiming for triple-digit revenue growth.
When considering a potential partnership and after identifying
employers’ key challenges, whether financial or otherwise, DTx entre-
preneurs must consider the path to monetization. Employers, unlike life
sciences players with more established monetization strategies, may
be willing to explore more creative DTx contracting approaches. T hese
include pay-per-use, subscription-based models, and outcomes-based
models. Employers who are open to new forms of collaboration typically
aim to shift away from focusing on return on investment (ROI) and
move t oward a value of investment (VOI) frame of mind. The value h ere
is the improved health and well-being of their employees.
DTx organizations should not forget that such partnerships can be
complex and, much as life science organizations view DTx technologies
as pathways t oward reenergizing patient-clinician engagement, employ-
ers view them as a tool to drive effective employee engagement with
their own health. For example, an overbearing, command-and-control
approach from the employer could negatively impact employee engage-
ment with the DTx platform, which in turn may affect the product’s
credibility through no fault of its own. Therefore, doing the necessary

groundwork to ensure organic buy-in, adoption, and adherence on the
part of both employers and employees are critical long-term success
factors.
On the part of employers, t here is also a lot of effort expected of
them in terms of properly assessing, implementing, and scaling such
interventions. To simplify this process, here are some tips for DTx in-
novators, based on learned experiences:
• Keep the buyers in mind when conveying the value of your

proposed solution. This may involve the human resources
function, which typically applies a risk-based approach to
assess which DTx platform to partner with.
• Go into meetings with potential employer clients prepared with
use cases and a business model ready for activation.
• Focus on a core set of solvable problems and d on’t stretch
yourself too thin.
• Keep it s imple: the targeted users are employees who may
already have access to other health-related platforms or tools.
Winning them over and achieving their engagement with a
newly introduced program is a key metric of successful adop-
tion and a prerequisite for value confirmation.
• Do your research to understand how the potential users work,
and how digital solutions could address some of the employers’
current challenges.
• Finally, keep in mind that employers frequently seek global
contracts. The scalability and adaptability of your solution and
value proposition approach across geographies may be impor
tant. This is not a trivial issue for behavioral-based DTx
platforms that tend to be highly context specific.
Final Thoughts on Employers

Employers may take some time to decide w hether and how to enter
into a partnership with a DTx provider, especially given the wide variety

of digital health offerings and the constraints of budget planning cy-
cles. However, they offer to DTx innovators the speed and reach that
few other channel partners do. As a majority of DTx innovation relies
on behavior change mechanisms that can have extensive and compre-
hensive effects at scale, DTx platforms serve as increasingly v iable
and effective solutions for employers seeking to support employee
health care and well-being.
Employers: A Partnership Example

The rise in chronic diseases in the United States, as in many other nations around
the world, has snowballed into a genuine crisis faced by the health care system.
More than half of US adults have at least one chronic disease, while over 27% have
been diagnosed with two or more chronic diseases.7 Diabetes, in particular, with
34 million affected adults, is estimated to cost the United States $367 billion per
year in treatment costs and productivity loss, not including the additional 88 million
adults who have prediabetes.8
Livongo developed a solution, a “personalized health program” as they call it, to
help manage diabetes, focusing on patient empowerment to improve health
outcomes and lower disease management costs. Specifically, it has developed a
health signals platform by combining a mobile app, personalized coaching,
connectivity to the cloud and other devices, and data science—providing a
comprehensive and clinically validated care delivery model. A peer-reviewed study
compared medical expenses of 2,261 Livongo program participants over a year to
8,741 matched individuals with diabetes who were not part of the program. The
Livongo diabetes program resulted in a 22% reduction in medical spending, which
translates into a US$88 saving per member per month at one year.9
In an important early decision that would propel Livongo’s scale-up journey, the
DTx company chose to focus its commercialization and channel partnership efforts
on employers. It did so in part because enabling employees to access tailored health
coaching programs resonated with its whole-person philosophy and platform to
support people, especially t hose with chronic medical needs, to live better and
healthier lives. It also made a strategic choice to support employers in keeping their
staff in top shape, aligning both personal and corporate well-being objectives.

Livongo’s partnership with Dean Foods is an example of how it works in
practice and what it means to an employer’s finances. Dean Foods had nearly
1,000 diabetic employees and faced medical costs in excess of $100,000 each
year. Within one year of integrating the Livongo platform into its employee wellness
programs, it achieved a 5% reduction in medical costs due to a 0.7 reduction in A1c
and a 28% reduction in hypoglycemia. Dean Foods’s return on investment for this
pilot program was 1.4 times—defined as cost savings from improved clinical
outcomes divided by the investment in Livongo’s solution.10 And the importance of
a strong partnership was clear from the get-go—Dean Foods championed Livongo
internally, observing an 86% activation rate among the targeted employees and a
net promoter score (a measure of customer experience) exceeding 71 (versus the
average of 12 obtained by corporate health plans). Not surprisingly, only a few
months later Livongo successfully raised a $45 million in a Series C funding round
and was acquired by Teladoc in 2020.11
Insurers: Potential Long-Term Strategic Partners

If there’s one part of the health care value chain that is being disrupted
more than any other, it is the private insurance market, particularly
in the United States and Asia. In the pursuit of differentiation and cus-
tomer loyalty, insurers are looking for ways to offer policyholders
value-added services (VAS)—additional features and solutions beyond
their core products. Medical second opinions, patient concierges, and
telemedicine consultations are examples of such serv ices. Recently in-
surers have expanded their offerings to include value-added wellness
serv ices, such as fitness trackers and customer rewards programs to
encourage physical activity. In Asia, for example, some insurance pol-
icies have offered as many as 40 value-added serv ices, including some
at early stages of clinical evaluation or with l imited validation of eco-
nomic utility to insurers. Of note, implementing value-added serv ices
with limited clinical evidence may result in health and life insurers in-
curring higher per-employee expenditures with no apparent benefit
to employees or employers.

Health insurance plans seem to be drawn to value-added serv ices
because many of these added features capitalize on current wellness
trends. The insurance industry is embracing this trend in part because
it is looking to improve its image and reverse the direction of its net
promoter scores that measure customer loyalty and satisfaction, a good
predictor of customer retention and business growth. Unfortunately,
these scores in the insurance industry tend to be negative, meaning
that more customers are “net detractors” rather than “net promoters.”
Customer experience is critical for health insurers b
ecause it is one
of the primary reasons why individuals renew or cancel their cover-
age. E
very time a member uses health serv ices, health insurers have
numerous opportunities to help the member, from the initial diagno-
sis to the payment claim. It is a l ittle more complicated for life insur-
ance. A generally healthy person w ill typically only have two to three
interactions with his or her life insurer per year. Too often, it may be
limited to annual policy renewals or regulatory revisions. Expectedly,
this can lead to disengagement and even irritation, especially if mem-
bers’ expectations for experience have been heightened through the
rewarding experiences in other areas of their lives (e.g., ride-hailing,
e-commerce platforms).
In a highly competitive business context, the pressure to balance
monetization with an expanded offering of serv ices that provide tan-
gible clinical benefits is creating a fertile ground for disruption. DTx
technologies represent a novel opportunity to insurers, beyond sim-
ply improving customer experiences. They can provide safe and effec-
tive solutions, improving health care outcomes for insurers’ consum-
ers and perhaps saving money. But how can insurers formalize this
type of value-added serv ice into a new business model?
One option for insurers is to provide their customers with a single
platform or app that contains a variety of VAS solutions, a sort of VAS
marketplace. VAS offerings have ranged from historical (e.g., medical
second opinion) to wellness (e.g., fitness and step c ounter apps), and
more recently, disease management (e.g., DTx for diabetes and cogni-
tive conditions). Customers, on the other hand, may become confused

if they are presented with multiple solutions from various vendors. If
a solution proves to be effective, clients may forget that their insur-
ance covers and frequently pays for these serv ices. If things don’t go
as planned, it is usually the insurer who is blamed.
Another option is to bring all VAS solutions u nder one brand, that
is, to rebrand u nder the insurer’s brand and deliver a single end-to-end
customer experience. Providing such a seamless customer experience
necessitates insurers developing new capabilities: designing back-end
systems, integrating data, managing liability and accountability, and
so on. Insurers are also still learning about the range of solutions that
customers expect from insurers or from independent providers, due
to concerns about data privacy.
Independent of the deployment options, if a DTx organization wants
to engage with an insurer as its distribution channel, an essential
factor to remember is that traditionally, insurance is a risk-averse in-
dustry, particularly when it comes to partnership. Consequently, the
near-term strategy for DTx solutions to be distributed as value-added
serv ices may target existing schemes and products that can be opti-
mized through a DTx intervention (e.g., delivering more efficient and
personalized health care, avoiding repeated claims from complications,
etc.). This strategy also provides the possibility of evaluating the DTx’s
cost-effectiveness, by comparing the claims before and after interven-
tion, at a scale based on a large user population. Therefore, an under-
standing of cost-effectiveness sets the stage for new collaboration
models, pay-for-outcomes, or value-based payment based on clinical
outcomes, thus moving away from the current fee-for-service reim-
bursement model.
Final Thoughts on Insurers

Partnering with insurers is a realistic option for scaling DTx deploy-
ment and should be pursued. Moving forward through this channel,
DTx w
ill be evaluated based on their influence on a specific client type

and coverage benefit scheme, with the goal of improving clinical and
financial outcomes as well as customer satisfaction. If a DTx partner-
ship is cost-effective, insurers may be more than a vendor to DTx
organizations; they could also be possible strategic partners in their
development, distribution, and commercialization.
Insurers: A Partnership Example

Cigna, a multinational insurer, and Omada, a pioneer in digital behavior change and
one of the largest providers of the Diabetes Prevention Program (DPP), have one of
the industry’s most well-known collaborations. Omada’s diabetes program uses
behavioral science techniques to assist members in managing their conditions with
the support of a virtual care team and connected devices, such as a blood glucose
meter or continuous glucose monitoring sensor.
Omada presented data from a randomized clinical trial involving almost 600
participants in 2020. One group was assigned to the Omada digital DPP program,
while the other group attended traditional in-person sessions. Participants in
Omada’s program observed an average reduction in HbA1c of 0.23% at 12 months,
compared to 0.15% in the comparison group. Also, Omada participants achieved an
average weight loss of 5.4% one year after enrollment, compared to 2% weight loss
in the comparison group. Overall, the study demonstrated that their digital diabetes
prevention program was as effective as in-person sessions.12
Cigna and Omada’s partnership started in 2015 when Omada began working with
just a few of Cigna’s self-insured employers. They have steadily broadened their
collaborations since then. As a nonexclusive partner, Cigna led Omada’s US$50 million
funding round in 2017. Cigna wanted to shift its business toward an outcomes-based
financing structure that aligned well with Omada’s pay-for-performance model.
During a pilot run in 2018 between the two partners and many large employer
clients, Omada’s solution enabled the participants’ sustained weight loss of up to
5%, a 9% drop in medical costs, and an overall ROI of 2:1. Participants also reduced
their risk for type 2 diabetes by 30%, a disease for which insurers typically pay
claims of US$17,000 per person per year.

TeleX Organizations: The New Front Door
TeleX, a catch-all term for telehealth, telemedicine, and telecare organ
izations, is widely seen as an entryway to modernizing the delivery of
health care and making it more convenient for both patients and phy-
sicians. Many health care systems are dated, inefficient, and not as
cost-effective as they could be due to gaps in care delivery. Their areas
for improvement include patchy coordination between levels of care
and challenges around patient access that are neither acceptable from
a customer experience standpoint, nor fit-for-purpose from outcomes-
based assessments. TeleX traditionally has three basic modalities:
synchronous and asynchronous patient-provider interaction, as well
as remote patient monitoring. It promises to improve patient ease and
access to health care, as well as quality of serv ice and patient satisfac-
tion. In short, telehealth promotes the development and deployment
of innovative health care delivery methods.
Despite the great potential and many successful teleX models, plat-
forms, and prototypes, there was substantial resistance to teleX
adoption prior to the COVID-19 pandemic. With the overburdening of
health care facilities and restrictions on physical access to health care
mediated by the coronavirus crisis, even the most tele-resistent pro-
viders warmed to the implementation of teleX solutions. In addition,
regulatory barriers have been coming down for telehealth, enabling
greater access and reimbursement. In the United States, 22 states
changed laws or policies during the pandemic to require more robust
insurance coverage of telemedicine.13 The Centers for Medicare & Med-
icaid Serv ices also expanded telehealth codes for the 2021 physician
fee schedule. Privacy concerns, for better or worse, are also relaxing
as populations are seeing the advantage of sharing data in return for
greater protection and health system navigation.
In 2021, telehealth utilization has been 38 times higher than before
the pandemic, which in turn has translated to a market share increase
from around 11% prepandemic to nearly 50% at present for teleX

platforms and services.14 According to a recent McKinsey & Company
report, doctors were able to see up to 175 times more patients than they
did before. Three out of four teleX users reported high satisfaction, and
over 50% of doctors reported having a dopted a more positive attitude
toward teleX since the pandemic began.15 Exemplifying this upward
trend for teleX solutions, Teladoc, a US telemedicine and virtual health
care company, had over 10.6 million visits in 2020 (+156% year on
year).16 In China, Ping An Good Doctor, a leading online health care
services platform, increased by 23.9% year on year their average daily
consultations, to 903,000 times per day.17 These figures confirm that
the response to the COVID-19 pandemic accelerated growth in tele-
health and virtualization of care pathways. An increasing number of
brick-and-mortar health care organizations made the shift to virtual
consultations, remote monitoring, personalized health records, and e-
prescriptions as part of their own digital transformation efforts. T hese
pre-and post-pandemic trends, together with the change in the regula-
tory attitude toward teleX, indicate that this important domain is
likely h ere to stay. With that, opportunities for creative partnership are
growing, too.
DTx and teleX companies are seeking to virtualize care pathways. Spe-
cifically, they are focused on delivering more person-centered care that is
closer in time and space to patient needs, collecting patient-reported and
biometric data that enable real-time insights for physicians, and so on. It
is worth exploring partnerships in which offerings are complementary.
For example, in 2020, Pear Therapeutics and UpScript Health, which pro-
vides a direct-to-consumer telemedicine platform, partnered to deliver
a solution for chronic insomnia. This integrated solution includes a
10-minute telemedicine consultation with a licensed UpScript doctor to
confirm if Somryst (the Pear Therapeutics DTx intervention) is appro-
priate for the patient. If so, the prescription is sent directly to the pa-
tient’s phone, allowing them to download and access the app. Patients
pay US$45 for the appointment with the doctor and, if they decide Som-
ryst is right for them, the cost of the program w ill be an out-of-pocket

expenditure.18 In this scenario, not only are Pear’s platform and Up-
Script complementary, but teleX is also a DTx enabler. (Before this
book’s publication, Pear Therapeutics began a reorganization.)
Another value proposition that teleX–DTx partnerships represent
is the opportunity to reach and enroll users that historically have been
out of reach for traditional health care players. reach52, a Singapore-
based social enterprise whose goal is to digitally reach the 52% of the
global population without health care, focusing exclusively on such last
mile, rural health access. As the company’s CEO, Edward Booty, shares
with us, “Most wealthy populations are on version 10+ of the Android
operating system. reach52 deals with communities and users on ver-
sion 6, which is over a decade old.”19
While reach52 has thus far served as a “platform of platforms” for
mainstream products to run health campaigns and offer access to med-
icines and insurance plans, the company envisions incorporating dig-
itally delivered therapeutics into its portfolio. It hopes this w
ill help it
overcome many of the supply chain, cost, and scalability challenges
startups like it typically face. Alongside its enterprising, socially re-
sponsible health care approach, the company also seeks to improve the
digital maturity of the workers and communities it reaches so that dig-
ital health solutions become more acceptable and adoptable.
Final Thoughts on TeleX

As health care progressively virtualizes, there are numerous opportuni-
ties for DTx and teleX companies to collaborate. DTx can leverage the
rising ecosystem adoption of teleX to reach a large user base. DTx pro-
vides clinically established solutions that extend beyond initial consul-
tations. TeleX companies can capitalize on this opportunity to offer new
products or services and build stronger relationships with patients.
Finally, teleX can be inspired by the clinical validation practices of
DTx, and DTx can learn from teleX about scalability through regula-
tion, clinician engagement, and technology integration considerations,
such as data protection.

TeleX: A Partnership Example
In August 2020, US-based diabetes management DTx company Livongo was
acquired for $18 billion by telehealth giant Teladoc. The union of the two companies
created a $38 billion global powerhouse that provides virtually hosted health
services in 175 countries across 450 medical subspecialties. Livongo has over 700
million data points on diabetes treatment trends, while Teladoc is the “world leader
in whole-person virtual care.”20
The deal was called the “biggest digital health deal ever”21 and a “transformational
opportunity to improve the delivery, access, and experience of healthcare for consum-
ers.”22 It offers two interesting opportunities. First and foremost is the opportunity it
represents for patients to experience the next generation of health care, including DTx.
It represents the possibility of a more data-driven and personalized approach,
providing a seamless consumer experience through the continuum of care.
Second, it opens a new cross-selling potential for both organizations. Initial
research indicated just 25% overlap between Livongo and Teladoc clients across
health plans, employers, and other channels. The potential to cross-sell in a
DTx–telehealth partnership outweighed the potential risks and enabled the
combined entity to be relevant to a broader user base with a more holistic solution
set. After completing the acquisition in just three months, Teladoc has announced
major deals, gaining momentum on cross-selling potential. While validation studies
and establishing evidence of clinical benefits are ongoing, analysts believe such
teleX–DTx initiatives will facilitate the provision of outcomes-based care that our
health systems need.
Telcos: A Long-Term Outlook

Wedged between teleX and Big Tech companies is an often overlooked
but highly relevant DTx channel partner: telecommunications compa-
nies (telcos). And they are everywhere—providing the connection
networks over which we communicate, the phones in our hands, in-
ternet access, content streaming, and so on. B ecause they are so ubiq-
uitous, telco players cannot escape the fate of any class of products or
serv ices that are readily available. They are converging on features,

quality, and price, which are eventually leading to a race to the bot-
tom in terms of profits. Faced with a downward trend in core reve-
nues, they are at risk of being seen as pure utilities players. So, to
avoid acquiring such a mundane image, and awakened to the realiza-
tion that they need new markets and revenue streams, many of them
are viewing the explosion of virtualized health care as an attractive
opportunity.
To make the most of that opportunity, telcos need to undergo their
own digital transformation first. In recent years, they have been in-
vesting heavily in developing and integrating 5G connectivity, cloud
platforms, artificial intelligence (AI) capabilities, and a host of digital
serv ices. Nine of the world’s largest telcos, including China Mobile, a
leading telecommunications services provider in mainland China, have
already adopted open application programming interfaces (APIs)—
made publicly available to software developers—for digital serv ice
management.23
Telcos tend to be regional brands with established levels of trust,
credibility, and government relations. They are also local employers
that are often summoned to support the development of informa-
tion technology infrastructure of health care systems and organ
izations. The general perception is that telcos, relative to Big Tech, are
more attentive to and compliant with data privacy and security regu-
lations, and this is definitely an advantage from the perspective the
health care industry.
We have seen this favorable predisposition play out for a few tele-
communications companies. In the United Kingdom over the last two
decades, the BT Group has established important partnerships with
the National Health Serv ice (NHS). T hese include the 10-year NHS
Spine contract to develop systems and software to support more than
899,000 registered users. In Australia in 2018, telco g iant Telstra se-
cured a $220 million government contract to build the National Can-
cer Screening Registry.24 During the COVID-19 pandemic in South
Korea, the country’s three major telcos—SK Telecom, KT Corp, and
LG—joined forces with the country’s health care industry to enable

contactless care that was strongly recommended as part of social dis-
tancing measures. Other telco–health care integration efforts in South
Korea have included personalized digital health management serv ices
over the cloud, patient meal archiving through mobile phone pictures,
and health trackers in nursing homes.25
As the experience in Korea shows, telcos can “host” entities in need
of agile infrastructure solutions by fostering a marketplace of inno-
vative service bundles and offering access to diff erent target segments:
individuals or organizations. This may mean that telcos are vendors
of infrastructure first and channel partners second.
What does this mean for DTx companies in search of channel part-
ners? While some DTx organizations may view telcos purely as B2C dis-
tributors (i.e., facilitating app downloads over the telco network),
others consider them strategically positioned to support a B2B2C, “sell-
with” approach to commercialization—that is, finding the right part-
ner with the right solution to better serve the needs of end users. For
example, b ecause telcos can use their networks to aid in the manage-
ment of patients with chronic disease by sending early warning sig-
nals to physicians or f amily members, conceptually they can play a key
role in preventing avoidable complications and keeping people out of
hospitals. Further, once a partnership is formed, telcos can leverage
their existing arrangements with corporate customers to raise aware-
ness about DTx solutions.
Regarding the commercial arrangement that is best suited to a DTx–
telco channel partnership, a subscription-based model for access to
the DTx is most likely to be the default choice for telcos. DTx innova-
tors whose commercial gains in such a partnership would necessarily
be tied to t hose of their counterpart must know that the fees telcos
can charge are often set by the government. Therefore, profits will have
an inherent limit to growth. Both sides must be flexible and inventive
in their commercial models. Telcos are generally trusted providers,
though this varies by country, so reputational risk in partnerships is
frequently a top decision criterion for them, which is often influenced
by having a validated business model and a solid vision for data play.

Final Thoughts on Telecos
Some telecom companies are just considering entering health care,
while o thers, such as KT Corp, are already t here. They have made the
necessary investments to secure a long-term presence in the health
care ecosystem. Telcos, like insurers, w ill need to demonstrate their
ability to add value to solutions and services that drive clinical and ef-
ficiency gains, such as t hose provided by DTx organizations. Syner-
gies between telcos and DTx developers, as well as the feasibility of
such partnerships, must be rigorously evaluated with due diligence and
understanding of the regulatory contexts and pathways to impact at
scale. In the short-to medium-term, telcos are probably better posi-
tioned to serve as an enabler to DTx organizations.
Telcos: A Partnership Example

Our case study here is the experience of Sibel Health, a US-based spin-off from
Northwestern University that specializes in biosensors used in intensive care units.
They received the Spinoff Prize26 for developing flexible devices to monitor heart
rate and blood pressure in premature babies without the need of adhesive patches,
which may harm a baby’s delicate skin. Using their sensor technology, they
launched a tracking and monitoring system of all essential vitals of COVID-19
patients in the United States and around the world.
If we are not in the DTx space, we can always learn from adjacent sectors and
solutions. You may be wondering why or how a medical device company like Sibel
would collaborate with and leverage any telco capabilities, but they did announce a
collaboration with KT Corp, South Korea’s largest telecommunications provider, in 2020.
KT aspires to be the leader in the crossing worlds of 5G, the Internet of Things
(IoT), and blockchain technologies, rather than just South Korea’s telecom leader. In
a nutshell, they want to be the world-leading digital platform company.27 KT was
eager to expand and provide additional customer services for COVID-19-related
needs. They chose to partner with Sibel to strengthen its capabilities in AI and big
data–based digital health. A pilot project is now underway to assess the viability of
deploying ICU-based remote monitoring services in developing countries.

KT had previously developed a Global Epidemic Prevention Platform, which alerts
travelers to epidemic outbreaks and provides relevant information while tracking
their movements to help prevent disease spread. In 2020, KT received funding from
the Bill and Melinda Gates Foundation for a Next Generation Surveillance Study for
Epidemic Preparedness, as well as a grant from the Korea-based RIGHT Fund to
assist in measuring the risk of COVID-19 infection.
With KT’s innovative infrastructures, and Sibel’s advanced biointegrated sensors,
both organizations want to “unlock the potential of AI and Big Data for the COVID-19
pandemic management and beyond.”28 The collaboration between KT and Sibel is
just the beginning. Telco companies are making a bigger push into health care, and
they’re seeking potential partners like established and aspiring DTx organizations to
quickly build capability and differentiation in the markets.
Big Tech: Potential Enablers

The opportunities for channel partnerships with diverse segments of
health care actors that we have discussed so far have hopefully piqued
DTx innovators’ curiosity and encouraged them to seriously consider
and weigh their options. Now let’s turn our attention to the elephant
in the room—Big Tech. It is no secret that Amazon, Apple, Facebook,
and Google have a ravenous appetite for getting into the health care
space, and their success in the customer experience game—coupled
with their enormous cash reserves—position them as serious players
over the coming decades. A wave of partnerships, acquisitions, and in-
frastructure build-out is already under way and likely to continue.
Kodak, the well-known photography brand, is an interesting exam-
ple of this trend. Some may refer to it as “the original Big Tech player.”
They invented and revolutionized photography. Unfortunately, Kodak’s
story and downfall is almost as well-known as their cameras. A fter
filing for bankruptcy protection in 2012 following 130 years of opera-
tions, the company saw its shares surge by 1,500% eight years later,
in July 2020, when it announced a strategic pivot into pharmaceuti
cals. Kodak w ill use its extensive technology experience in chemicals

and materials manufacturing to produce active pharmaceuticals in-
gredients, of which only 10% are manufactured in the United States
today.29 Kodak’s story demonstrates the hidden opportunities the
health care industry holds for established but imaginative technology
brands that are willing to leverage their strengths in new ways.
In Asia, groups such as AliHealth (a subsidiary of Alibaba) and Ping
An Good Doctor (a spin-off of Ping An) are following a similar trajec-
tory. Ping An, originally a traditional insurance company, is widely
known t oday as a financial tech and health tech player. It has over
US$200 billion in cash reserves (more than double those of Amazon),
over 200 million retail customers, and over 500 million internet us-
ers.30 The company reinvests 1% of its revenues into research and de-
velopment to expand its technology businesses, particularly in the ar-
eas of AI, blockchain, and cloud computing. Ping An Good Doctor, the
one-stop-shop health care serv ices platform of Ping An, has close to
70 million monthly active users who account for approximately
700,000 daily telehealth consultations.31 When the company went pub-
lic on the Hong Kong stock exchange in 2019, it raised US$1.1 billion.
Its business model, based on one customer, one account, multiple prod-
ucts, and a one-stop shopping platform appears to be its recipe for
scalability—other tech players likely have their own r ecipes and se-
cret sauces too.32
In a sense, Big Tech companies represent a gateway to other chan-
nel partners rather than standalone partners themselves, so in order
for DTx innovators to capitalize on the market power they bring, it is
important that they understand the most common business models
under which Big Tech operates. They don’t tend to copromote any par
ticular solution on their platforms but instead facilitate connections
between potential customers and potential solutions. This is in part
because the Big Tech industry has been built around the philosophy
of “self-serve,” which on the one hand aligns with the general vision
of DTx innovators in terms of being self-driven. On the other hand,
Big Tech behemoths are mostly interested in being in the business of
cloud and data management, so they are often agnostic to the ratio-

nale of particular technologies that make use of their infrastructure.
Big Tech can thus help DTx ventures with such strategies as setting
up architecture in a cost-efficient manner and accommodating local-
ization needs, such as data compliance and language processing.
Many of the Big Tech players offer marketplaces too. DTx solutions
can be presented in t hese settings to drive searchability as well as in-
teraction with the user community. In addition, t hese solutions can
be paid for based on pay-per-use, pay-by-payers, or pay-by-employer
models. Amazon, for example, has already seen a rise in genomics-
related offerings and disease algorithms being managed on Amazon
Web Serv ices. This setup has certain advantages compared to DTx col-
laborations with more traditional partners, such as pharma compa-
nies that require an advanced level of clinical validation to be com-
pleted on both sides before engaging in discussions. On the other
hand, Big Tech organizations tend to follow a shared responsibility
model in which they guarantee data security while the DTx venture
takes ownership of data validity.
Big Tech’s forays into health care h aven’t all been successful. Ama-
zon’s first foray into the health sector was a 1999 investment in Drug-
store.com, an online retailer of health and beauty products. Although
Amazon was ahead of its time, e-prescriptions and e-pharmacies w ere
still in their infancy and the venture was not successful. For the 2018
fiscal year, Amazon collaborated with Berkshire Hathaway and JPMor-
gan Chase to launch the venture Haven. Three years later, the company
that was founded to disrupt US health care was shut down. The com-
bined 1.2 million employees of the founding companies were insuffi-
cient to wring lower prices from providers and shift to a capitation
model—a fixed amount paid to physicians for each enrolled person, per
period of time, whether or not that person seeks care. In addition, the
company’s internal priorities diverged, causing it to close in early 2021.
Amazon is not the first company to fail in its attempt to disrupt
health care. Remember that Google announced in 2011 that it would
discontinue its personal health record platform due to scaling issues.
Google’s health division was then dismantled 10 years later, in 2021.

The same week, Apple confirmed that one of its most ambitious health
care projects, HealthHabit, an internal application that Apple employ-
ees can use to track fitness goals, communicate with clinicians, and
manage hypertension, would be scaled back. Big Tech companies are
not omnipotent. Teaming up with these prominent and well-established
institutions does not guarantee the venture’s success.
Final Thoughts on Big Tech

Tech g iants have made substantial contributions t oward modernizing
the health care industry. However, our previous examples have shown
that the pathway between ideation and adoption can be fraught with
obstacles. Disruptive technologies are insufficient on their own to dis-
rupt a complex and fragmented industry like health care. To solve pa-
tient problems in health care, unique domain expertise that is seam-
lessly integrated is required, with a focus on the patient’s experience
rather than solely the validation of the technology. This is where DTx
organizations can contribute, by leveraging emerging technologies as
foundations for implementation, adoption, and scalability.
W hether it be incumbent or emerging Big Tech players, their moti-
vations, financial resources, and infrastructural breadth and depth
still represent an attractive channel partnership opportunity for DTx
innovators. As DTx entrepreneurs ponder such collaborations, how-
ever, they should remain mindful of how to best pair an emerging
tech platform with the established infrastructures and resources of
large tech organizations in order to best position both parties for a
win-w in outcome.
Big Tech: A Partnership Example

As discussed in this section, the mechanism of a DTx–Big Tech partnership is often
one of Big Tech companies offering the infrastructure, platform, and data capabili-
ties necessary to propel a DTx or startup DTx innovation to greater scalability. This
was the case for US-and Singapore-based Savonix, a neurocognitive assessment

mobile app used to screen, track, and evaluate brain health, particularly as it
pertains to possible early-stage impairments related to Alzheimer’s disease and
other cognitive health conditions.
Inspired by the Apple Heart Study, which was set up in 2017 in collaboration with
Stanford University to analyze heart data recorded by Apple Watches and iPhones
and identify irregular heart rhythms, in 2019 Boston University School of Public
Health partnered with Savonix to conduct a first-of-its-kind population health study
in dementia, known as the Alzheimer’s Disease Discovery Study (ASSIST). The study
sought to digitally collect health and lifestyle data from 400,000 people, asked
them to complete a short health history questionnaire, and administered a Savonix
Mobile cognitive test to assess cognitive function. The collaboration’s goal was to
identify differences at the personal level that could lead to the development of
future interventions and treatments, such as lifestyle interventions that could be
used to better prevent or manage Alzheimer’s disease, in addition to genetic factors.
For the study, no doctor visits or brain scans were needed, but study partici-
pants were required to have an iPhone in order to be included in the study. They
were enrolled online and asked to download the Savonix mobile app and take a
series of its cognitive tests. In addition, they were asked to share exercise,
nutrition, and sleep data from the iPhone Health app.
This collaboration was enabled by participation from Apple. The clinical outcome
of the partnership is not known yet, but the potential is high it can lead to the
development of targeted therapies for Alzheimer’s disease by looking holistically
across population demographics and multidimensional health factors.33 Of note,
Savonix didn’t stop t here and is keeping the Big Tech channel partnerships moving
while the ASSIST study is recruiting participants. In 2020, they announced a
partnership with Fujitsu Connected Technologies Limited, one of Japan’s largest
smartphone manufacturers. Savonix Mobile, the digital cognitive testing app, has
been preinstalled on over a million of their Raku-Raku smartphones. The phone also
includes a pedometer and other important preinstalled health care applications to
support and manage the users’ health.
The case study of Savonix shows how tech giants can act as an enabler, like
Apple, as well as strategic partner, like Fujitsu in this case, in the distribution of
digital health and DTx solutions.34

Conclusion
The timing has never been better for DTx innovators to seek out chan-
nel partners as a path to commercialization. While such collaborations
are not required for DTx companies to succeed, the challenges associ-
ated with going the commercialization journey alone can be tough, and
even insurmountable. Hence, we would suggest evaluating the likely
upsides and risks involved in taking the partnership route on the road
to wide-scale adoption. The good news is that the list of prospective
channel partners is long and growing, given the keen interest in the
health care sector by various market players expanded by the COVID-19
pandemic. From pharma to telcos to insurers, as well as newly emerging
potential players, global enthusiasm toward innovation in the health
care industry is at an all-time high. There are a variety of channel part-
ner types, with respective strengths and weaknesses, for DTx innova-
tors to explore. Or go it alone, should the circumstances allow.
Having a multipronged commercialization strategy and being pre-
pared to pivot swiftly will probably be needed, especially in an indus-
try such as health care, which is frequently being disrupted. As dis-
cussed earlier in the book, the most important element is keeping the
problem your DTx solution is addressing as your true north during
such twists. As long as decisions and course-corrections do not deviate
your team from solving an established challenge, it is likely to be the
most direct way to success. In short, remember to stay centered for the
journey ahead.
As far as developing DTx technologies that do solve health-related
challenges, we also cannot emphasize enough the importance for in-
novators to keep a steady focus on the heart of the m atter: clinician,
patient, and caregiver engagement. Many DTx innovators will have had
some level of interactions with clinicians during concept validation
and prototyping, but as they advance through the later stages of se-
curing a channel partner and getting their product on the market, they
may lose sight of the user experience—and that would be a mistake.

Maintaining a positive user-DTx experience—the “last mile” from a
care journey and health systems perspective—is where the different
strands of the DTx go-to-market journey merge together. Clinical val-
idation, commercialization, and channel partnerships must ultimately
all lead to that goal.
Just as t hese potential partners can be expected to scrutinize DTx
solutions for commercial viability, so too should DTx innovators care-
fully assess their respective go-to-market channels with a view to
achieving market longevity and sustainability. With so much at stake
on both sides, it is understandable why such mutual “business part-
nership due diligence” must be performed. Based on our interviews
with experts collaborating with DTx organizations, a summary of the
key factors that DTx organizations would do well to consider when
evaluating a potential channel partnership is found in table 11.2.
T able 11.2. Key factors to consider as part of DTx’s business partnership

due diligence
Strategic fit Operational fit
• Is the partner’s portfolio of offerings and data • Can the partner provide the necessary
complementary to yours? front-office resources (e.g., sales, marketing,
• Is t here an alignment with the partner in the access, pricing, regulatory, and medical)?
markets (demographic or geographic) being • Can the partner provide the necessary
targeted? supply and distribution channels?
• Is t here a clear, mutually aligned, and • Is the partner able to influence end users,
measurable definition of success for clinicians, and administrators? In other
partnership? words, does it have sufficient experience in
health care or in the therapeutic area of
interest to the DTx venture?
Financial fit Cultural fit
• Is t here alignment on the monetization model • Do both parties share a common vision and
and flexibility therein? mission for health care?
• Does the partner have access to sufficient • Does the partner have an acceptable risk and
capital resources and incentives? timeline tolerance?
• Are both parties clear on how the partnership • Does the partner place emphasis on people
could evolve (e.g., mergers and acquisitions)? and innovation development?

To support DTx innovators in making the right choice when select-
ing channel partners, and attending to the factors outlined in
table 11.2, we leave you with the following recommendations:
• Choose the right partners. Perform your own due diligence on

the readiness, willingness, abilities, and cultural fit of the
prospective collaborators.
• Conduct discovery. Explore what matters to potential channel
partners in terms of priorities, growth, unique selling proposi-
tions, challenges to overcome, and customer base.
• Set goals and commit. Create a plan with objectives, target
markets, strategy, roles and responsibilities, resourcing, and a
realistic ROI.
• Facilitate introductions. Invest in the relationship, assign a
dedicated partner manager, and ensure continual feedback
about the latest product and market developments.
• Enable the partners. Provide technical product training, relevant
domain expertise, and sales guidance, paying special attention to
those partners contributing toward a majority of the partnership.
• Support the sales, transparently. Ensure that pipelines and
pitching opportunities are shared between partners, jointly
overcoming any roadblocks and embedding customer success as
a continuous improvement activity.
• Evolve the program. Conduct and monitor surveys, net promotor
scores, reflections, and regular joint strategic planning sessions
to stay aligned and ensure focus on the right set of activated
channel partners.
Seek out win-w in partnerships—potentially transformational for

not only DTx organizations but also all partners and ecosystem stake-
holders, including patients, and especially those channels that are
likely to result in long-term change in health care practice and deliv-
ery. This is especially true at the time of accelerated growth of digital
health, when there is a bit of a gold rush for digital applications in health
care; for DTx startups and their channel partners to make a difference,

the entire ecosystem (patients, clinicians, policymakers, regulators)
must be aligned for the ride ahead.
What can innovative DTx companies expect in return for all this
strategic evaluation and partnering? Accelerated growth, greater brand
awareness, expansion of revenue potential, and new markets are all
potential outcomes. We explore how to effectively articulate the value
of your DTx in the next chapter.
Key Takeaways
• As DTx innovators prepare to take their solution to market, the scalability and
sustainability of the company’s commercialization efforts will likely require
channel partnerships.
• There is a variety of channel partnership models available, traditional as well
as novel, and companies must analyze which modality is most compatible
with their product, end user, and business objectives.
• While DTx is a new and evolving space within health care, there are already
sufficient real-world examples of both successes and failures that can serve
as guideposts for entrepreneurs and help them avoid costly mistakes.
Notes
1. Staton, “The Top 10 Patent Losses of 2015.”
2. Sanofi, “Sanofi and Abbott Partner to Integrate Glucose Sensing and Insulin
Delivery Technologies to Help Change the Way Diabetes Is Managed.”
3. Singapore News Center, “Zuellig Pharma Shapes the Future of Healthcare with
the Cloud.”
4. Akili Interactive, “Akili and Shionogi Announce Strategic Partnership to
Develop and Commercialize Digital Therapeutics in Key Asian Markets.”
5. Organisation for Economic Co-operation and Development, “Employment
Rate.”
6. Centers for Disease Control and Prevention, “Workplace Health Promotion:
Depression Evaluation Measures.”
7. Boersma, Black, and Ward, “Prevalence of Multiple Chronic Conditions among
US Adults, 2018.”
8. Centers for Disease Control and Prevention, National Diabetes Statistics
Report 2020: Estimates of Diabetes and Its Burden in the United States.
9. Whaley et al., “Reduced Medical Spending Associated with Increased Use of a
Remote Diabetes Management Program and Lower Mean Blood Glucose Values.”

10. Dean Foods and Livongo Health, Livongo Drives 1.4x ROI in Year 1 for Dean Foods.
11. Landi, “Telehealth Leader Teladoc to Buy Livongo in $18.5B Deal.”
12. Omada Health, “Omada Digital Diabetes Prevention Program Shows
Sustained HbA1c Reduction and Weight Loss in Randomized Control Trial.”
13. Volk et al., “States’ Actions to Expand Telemedicine Access during COVID-19
and F
uture Policy Considerations.”
14. Bestsennyy et al., “Telehealth: A Quarter-Trillion-Dollar Post-COVID-19
Real ity?”
15. Bestsennyy et al., “Telehealth.”
16. Landi, “Teladoc’s Virtual Visits Reach 3M during Q4 as Revenue Grows to
$383M.”
17. Ping An, 2020 Annual Report, 65.
18. UpScripts Somryst, “Train your Brain for Better Sleep.”
19. From an interview with the authors, September 17, 2021.
20. Teladoc Health, “Teladoc Health to Announce Third Quarter 2022 Financial
Results.”
21. Forrester, “The Biggest Digital Health Deal of All Time: Teladoc and Livongo.”
22. Teladoc Health, “Teladoc Health and Livongo Merge to Create New Standard
in Global Healthcare Delivery, Access and Experience.”
23. TM Forum, “Telco Cloud Orchestration Plus, Using Open API on IoT.”
24. STL Partners, “Four Strategies for Telcos in Healthcare.”
25. Yonhap, “Telcos Expand Digital Health Care Serv ices amid Pandemic.”
26. “Scailyte among the Spinoff Finalists for the Global Nature Research Awards,”
Swiss Biotech.
27. KT Corporation, “S.Korea’s KT Corp. to Expand Digital Health Business with
Sibel.”
28. Koh, “KT Signs MoU with Sibel Health to Expand Its Digital Health
Capabilities.”
29. Blankenship, “Camera Maker Kodak Dives into Drug Manufacturing with
$765M Federal Loan.”
30. Ping An, Announcement of Unaudited Results for the Three Months Ended
March 31, 2020.
31. Ping An, “Ping An Good Doctor Has Become the First Online Healthcare
Platform with More Than 300 Million Registered Users.”
32. Ping An, “Ping An Unveils Health Care Ecosystem Strategy.”
33. Savonix, “Savonix and Boston University School of Public Health Launch
Landmark Population Health Alzheimer’s Disease Discovery Study (ASSIST) to
Digitally Collect Brain Health Data.”
34. Savonix, “Savonix and Fujitsu Connected Technologies L imited Partner to
Introduce Savonix Mobile App on Raku-R aku Smartphone F-42A.”

• Chapter 12
The Price Is Right

Value in the Eye of the Beholder
As we have seen through the multitude of real-world examples pro-

vided so far in this book, the health care industry is slowly pivoting
toward a value-based model in which reimbursement is increasingly
tied to patient health outcomes rather than to the volume of products
and services delivered. DTx technologies and companies are well-placed
to accelerate this shift through better data collection and cost-effective,
consumer-f riendly platforms. For this reason, DTx innovators must
have a pricing strategy in place that serves not only to engage poten-
tial channel partners but also to reflect the role their solutions play in
bringing legacy health systems and operations up to speed with evolved
expectations around value-based care. The question of pricing, there-
fore, lies in successfully conveying the value they bring to t hose part-
nerships and to the payer.
DTx monetization, especially in relation to reimbursement, is a new
and complicated topic. It is considered one of the key impediments to
DTx adoption. Addressing a pathway to reimbursement, if applicable,
should be considered early on. Not only must DTx entrepreneurs take
the time to understand the motivations of their potential channel
partners, but they should also identify payers and their dynamics in
231

Table 12.1. DTx commercial models in the United States
Minimum requirements
Archetype Reimbursement FDA clearance Prescription Clinical evidence
Business to Cash pay No No No (highly

consumer recommended)
Business to Providers (hospitals, No No Usually (highly
business practices), recommended)
pharma, or medtech
Business to Programmatic No No Usually (highly
payer spending recommended)
Claims-based Yes Yes Yes
reimbursement
Source: Hoult, “Key F actors to Obtaining Reimbursement for Digital Therapeutics in the US.”
the health care industry. Payers, which may be public or private, are
entities that w ill cover the cost of the serv ice provided by the DTx to
the end users. DTx innovators should explore all possible reimburse-
ment pathways that may be available to them through both types of
the payers while also seeking to understand the range of engagement
models for each payer (table 12.1).
The first step t oward developing a value-based offering is to define
what “value” is and how it w ill be measured. DTx organizations should
collaborate with their channel partners to reach an alignment in this
area as each stakeholder has its own incentives and metrics by which
it assesses value and makes decisions. DTx platforms that simulta
neously satisfy multiple parties have a higher chance of achieving
commercial success.
Providers typically consider something to be of value based on the
appropriateness of care and effective, evidence-based interventions.
Meanwhile, patients value health services or interventions if they lead
to clinical improvement and better functional health and provide user
experiences that are as seamless and minimally invasive (both surgi-
cally and practically) as possible with minimal or no disruption to their
personal, professional, and social lives. From health care managers’
perspectives, value could also be defined as the clinical benefits and

health improvements achieved for the money spent. To paraphrase a
familiar saying, value is in the eye of the beholder.
Which of these value definitions should be the focus of a DTx offer-
ing? The answer depends on the deployment context and channel part-
ners and may incorporate several value definitions tailored to the
business model. Let us take a closer look at the payers as well as some
potential f actors to consider when determining DTx value.
Public Payers: A Focus on the Evidence

Most health care systems around the world are publicly financed in
some way, meaning that reimbursement through the state is or should
be a key consideration for any DTx commercial strategy. Public payers’
awareness of DTx has grown significantly in recent years. Some initially
provided access to digital health apps and DTx as part of programmatic
spending for wellness and disease management. Others have been slow
to adopt new technologies or adapt their payment methodologies to
cover DTx. Many public payers now have or are thinking about estab-
lishing a digital health formulary. Regardless of the country where DTx
organizations operate, and the archetype of commercial models, public
payers w ill consider regulatory submission, clinical data, and health
economics outcomes for potential coverage as well as the payment level
at which they will reimburse. Put it simply, DTx organizations must
demonstrate that their products are medically necessary.
While going the fully validated, regulated route to commercializa-
tion may seem like a daunting, time-consuming, and costly exercise
for DTx innovators, this is the way translational science—the process
of applying knowledge and insights from basic science and clinical
trials to techniques, tools, and interventions that address critical
medical needs—operates. Robustly addressing this pathway differen-
tiates DTx platforms from nonregulated digital health solutions. This
is a must for the “golden ticket”: securing claims-based reimburse-
ment at scale. Arguably, it is one of the most difficult paths to achieve
in the context of bringing a product to market, since the arduous and
The Price Is Right 233

necessary process of clinical validation alone may not be sufficient for
commercial success. More specifically, this space also requires DTx
organizations to spend considerable time and effort demonstrating
the economic need for and efficacy of new treatments, and obtaining
robust evidence to support such claims is frequently done on a l imited
budget.
Most importantly, DTx innovators must be aware of the chasm be-
tween approval and reimbursement. Regulatory clearance, such as ap-
proval from the Food and Drug Administration (FDA), does not guar-
antee that any payor will provide reimbursement. Data needed to clear
the FDA are often insufficient for coverage and market adoption. T here
is a long list of novel medical technologies that are in “funding purga-
tory,” which means they have been approved as clinically sound but do
not have a clear coverage model in terms of who pays for them. This
again underscores the importance of having a reimbursement road
map as part of a DTx company’s go-to-market strategy. Some may ar-
gue that coverage pathways should be considered only if and when
marketing approval is obtained. In our view, given how important this
aspect is for DTx sustainability and integration into the broader health
system, formulating a vision and an action plan as soon as feasibly pos
sible is a differentiating advantage for serious innovators. That is why
clinical studies should be designed not just to demonstrate the safety
and efficacy of a solution but also to address payers’ potential concerns.
Payers are not only asking for outcomes data but long-term data as
well. What would be the long-term value and impact of a 12-week ex-
ercise and cognitive behavioral therapy intervention program, for ex-
ample? Would such digital intervention provide long-term benefits to
payers? Will these benefits “stick”?
Generating data, specifically real-world evidence, needs to be a top
priority. While focusing on public payers, we encourage DTx organ
izations to explore ways they can collaborate, early, with governmental
agencies and organizations, not only to obtain an endorsement but also
to collect real-world data. Singapore, for example, recently rolled out
mindline.sg, a one-stop mental health platform that matches users with

resources and helplines. Within that platform, it integrated Wysa, an
aspiring DTx that operates an emotionally intelligent chatbot that “lis-
tens” to people’s mental health struggles and anxieties and guides them
to appropriate cognitive behavioral therapy or coaching exercises.
The Wysa model is illuminating because while its success relies on
the government’s endorsement and adoption, the company focused on
gaining traction and business funding through a direct-to-consumer
model first before they approached the public sector for scale. Singa-
pore’s government currently pays for the Wysa license at a discounted
rate and offers the platform for f ree to people in the country. Outside
of Singapore, and perhaps eyeing a similar B2C-to-nationwide scale-up,
Wysa is also being deployed at Cincinnati Children’s Hospital in the
United States and at the North East London National Health System
Foundation Trust in the United Kingdom.
Singapore’s experience with incorporating digitally delivered mental
wellness services shows that if DTx innovators aspire to broaden market
adoption, they should consider what it would take to convince large
payers, such as central governments or regional health boards, to adopt
their solutions. Yet, all too often, DTx business and pricing propositions
are based on the founders’ and developers’ desire to recoup their research
and development investment rather than on a strong health economic
model that demonstrates the potential of their technology to achieve
cost savings based on improvements in patient outcomes.
Public payers cannot afford to overinvest. They may be unconvinced
by revenue-prioritizing commercial models u nless they clearly map out
their value to the health system. To stand a chance of successfully gar-
nering public reimbursement of their products, therefore, DTx inno-
vators must uncover these value-return ratios and extrapolate them
to the population level, showing how their solution alleviates illness,
disease management, or economic and other resource allocation–
related challenges. This stresses, once again, the importance of strong
health economics and outcomes research to bolster a DTx product’s
pathway to move beyond technical and clinical validation and achieve
sustained adoption.

Seeking public reimbursement is an important strategy for DTx in-
novators looking to scale. With this in mind and the points just men-
tioned, here are a few practical tips when approaching the public sec-
tor as a potential payer:
• Target major health care utilization shifts, such as the transi-

tioning away from care delivery to health preservation, from
hospitals to home, or from a focus on quality alone to the
overall value that is in line with the governments’ broader
strategic transformation goals.
• Adapt DTx solutions (or have a plan to do so) to the local context.
Governments are open to ideas from around the world, but DTx
innovators must show a clear capability for localization.
• Demonstrate value not only in financial or clinical terms but
also in terms of the DTx’s staying power, or “stickiness,” month
after month—defined by how memorable, valuable and habit-
forming a DTx is for any individual user so they w ill actually
continue to use the solution on a regular basis.
• Think about the potential business models (including reim-
bursement) and the scalability of your DTx on a national scale
early on (which is the perk of partnering with government).
Private Payers: A Focus on Returns and Value

In the aim of better price points, targeting private payers (e.g., in
a business-to-business approach or programmatic spending; see
table 12.1) tends to be the preferred option for DTx and aspiring DTx
solutions. DTx entrepreneurs can in fact negotiate on a case-by-case
basis with health insurance plans and employers. However, the potential
end-user base is usually smaller than that available through nationwide
coverage.
In many countries, insurance and employer benefits schemes are
being revised and regulated more rigorously than in the past, both by
the country or local governments and by the organizations themselves,

as payers attempt to assure adequate coverage while keeping overall
costs u
nder control. As a result, private payers are increasingly offering
their members health services and solutions that fall within a certain
financial threshold and are therefore assessing the “price asked” by the
providers of t hose services against the clinical and economic value they
generate, as we discussed e arlier. Furthermore, many private payers im-
pose co-payments, deductibles, and other out-of-pocket expenditures,
implying that not e very service included in an insurance policy is neces-
sarily fully covered and patients themselves share in the cost of care.
Because of the variability across these options in the private payer
market, defining a pricing strategy for a DTx product is more complex
than putting a price tag on a new medical device or therapy. In the ab-
sence of real DTx alternatives and of well-defined reimbursement
mechanisms, DTx innovators, should resources allow, consider pursu-
ing two avenues concurrently:
• Focus on value. DTx solutions that create measurable value for

patients, providers, and payers should be priced accordingly.
This includes understanding and measuring both users’ percep-
tions of the DTx’s benefits and the value it creates within and
along the care continuum.
• Focus on cost (and so, returns). DTx entrepreneurs should be
considering all the direct and indirect costs involved in devel-
oping, deploying, and distributing a DTx to their partner and
end users when assessing their cost-effectiveness.
ere we turn our attention to how DTx can align their pricing models
H
with the value they provide to patients, other health care stakehold-
ers, and the health care system in general.
Users’ Willingness to Pay

Understanding the value of a DTx technology starts with understand-
ing its target users’ willingness to pay (WTP) for the benefit it offers.
This is particularly relevant in contexts where providing such solutions

implies out-of-pocket expenditures. Households in countries that are
part of the Organisation for Economic Co-operation and Development
finance about 25% of the cost of prescription medication.1 In the United
States, it is common for patients to have a 20% co-pay or coinsurance.
In these situations, a DTx pricing model must account for a key wild
card—users’ willingness to not only use the solution but also to pay
for it. This f actor is also likely to influence its uptake by providers. (If
patients are not willing to pay for the solution, doctors w ill have l ittle
reason to prescribe it.)
Estimating potential DTx users’ willingness to cover the costs of use
is an important exercise DTx innovators need to undertake when de-
veloping their pricing strategy. It reveals the economic constraints
within which patients are willing to consider a solution. This impor
tant exercise is truly about understanding the real value-drivers of a
DTx to each cohort or customer segment, as well as t hese individuals’
willingness to forego the intervention in place of another solution or
other pressing personal expenses.2
A WTP analysis should also be entirely contextualized to the ge-
ography, socioeconomic status, and disease pathways of the targeted
patient population. It is important that this analysis addresses the
foundation of the solution—t hat is, what is the unmet need being
solved with the solution and what impact is the solution expected to
have? Such analysis is what drives the business case for channel part-
ners and payers to engage with DTx innovators who are keen on build-
ing up their portfolio of platform play capabilities. DTx entrepreneurs
should consider average income levels, discretionary spending, and
health or social balance for productive living to arrive at a figure that
is acceptable to the population while also remaining economically
v iable for DTx organizations and potential channel partners.
In a US study, for example, researchers studied participants’ willing-
ness to pay to be part of a blood pressure control study. They asked pa-
tients a series of questions about time and travel costs, as well as time
value related to their study participation. They found out that the
mean WTP was of $25 out of pocket for a doctor’s consultation.3 The

final pricing, in our opinion, is just as important as the factors that led
to the $25 price point, since it was discovered that more than 50% of
patients did not have insurance coverage or a designated primary care
provider. Based on findings like t hese, we also encourage DTx entrepre-
neurs to conduct a price elasticity analysis to pinpoint whether an in-
crease or decrease in prices would generate different outcomes, such as
a change in DTx consumption. How might a $30 versus a $20 consulta-
tion fee affect patients’ willingness to participate in the blood pressure
study and see their doctor? The same reasoning applies to DTx.
In summary, out-of-pocket payments strongly influence patient
decision-making as well as physician choice when it comes to prescrib-
ing a DTx solution. W hether your DTx commercialization strategy tar-
gets public or private payers (or both), channel partners will expect a
WTP analysis as part of any comprehensive distribution negotiation.
This is a reminder that beyond setting a price point for their product,
DTx innovators must first and foremost consider the holistic value
they are offering to each of the key decision-makers involved in its
uptake.
Pricing Models
When pricing their solution, innovators sometimes mistakenly believe
that mimicking a pricing model for another technology-driven niche
of the market, however successful (e.g., software as a service, or SaaS),4
w ill also catapult this new innovation to commercial success. Trans-
porting pricing and sales models from other sectors wholesale, how-
ever, rarely works in health care b ecause of the myriad unique biological
and organizational factors that characterize it.
In the course of our research for this book, we came across a vari-
ety of pricing model archetypes that could serve as an inspiration for
DTx innovators in their discussions with private channel partners.
They each have their merits relative to the industries and business con-
texts in which they are commonly used and, as mentioned, must be
tailored not only to health care’s singular context but also to the channel

T able 12.2. Popular pricing model archetypes adaptable for DTx, especially
relevant in a B2C and B2B business models
One-o ff payment Device and consumables
• Charge per intervention, procedure, test, or • “Razor-razorblade” model: relevant for

activity with reimbursement and/or solutions with consumables and/or such
out-of-pocket payment supplies as DTx for diabetes management
coming with strips, continuous glucose
monitoring sensors, and/or infusion sets or
accessories for the insulin pump
Subscription Maintenance
• Pure: subscription-based, including fixed • License: annual renewal, including technical

and variable components support and software updates
• Freemium: f ree basic service with optional • Rental: applicable for hardware, on a lease with
fees for access to advanced features depreciation costing
• Leveling: tiered customer categories based
on user profiles, utilization, and features
partners and payers with whom DTx entrepreneurs are considering to

engage on the path to commercialization. To help orient them in this
process, we highlighted a few of the most popular models (table 12.2)
and a list of questions whose answers should guide them in making
the right choices.
Checklist of considerations for selecting the right pricing model:
❒ What are your geography, therapeutic area, and clinical and

health economics outcomes data?
❒ Is the solution purely a digital offering or does it involve a
companion drug, medical device, or other equipment? If so,
how would DTx pricing fit into the reimbursement model of
the item or device that is now reimbursed or seeks
reimbursement?
❒ What is the current level of competition within the therapeu-
tic area being addressed? Are t here digital and nondigital
options? Would the DTx potentially replace an existing item
or serv ice already with a high reimbursement rate?

❒ What is the differential advantage of the proposed DTx com-
pared to what is already out on the market, both digital and
nondigital?
❒ What are the potential applicable reimbursement models or
archetype models as discussed earlier?
❒ Who is expected to pay for the product and the potential
co-payment/out-of-pocket expenditures for patients?
❒ Who are the targeted end users and what is their willingness
to pay if they are expected to bear all or part of the cost?
❒ What is the level of maintenance or servicing required to
maintain a current iteration of the solution? How regularly
does the software need to be updated?
❒ What could be the number of potential DTx prescriptions at
scale?
❒ How often are patients and caregivers expected to use the
DTx and for how long?
A combination of pricing strategies may often be appropriate, de-

pending on the therapeutic area being addressed, the type of business
model (B2B, B2B2C, or B2C), and the geography. For example, Blue-
Star® (the diabetes management solution powered by Welldoc that
was discussed in the earlier chapters of this book) offers a variety of
products and their respective pricing models to diff erent customers
(i.e., employers, health plans, or health systems) in a B2B2C approach.
JOGO, a US-based DTx company with international operations in Asia,
leveraging wearable sensors, artificial intelligence, and virtual reality
to treat neuromuscular conditions using neuroplasticity and biofeed-
back, is also a good example of business model and revenue diversifi-
cation. The company’s long-term strategy is to create a current proce-
dural terminology code (a set of medical codes that are used by health
care professionals to bill medical serv ices and procedures to public
and private health insurance programs for potential reimbursement)
so JOGO can be prescribed and reimbursed. Meanwhile it is explor-
ing three distinct revenues streams:

• DTx as a service. In the United States, JOGO is FDA-registered
with its service covered by Medicare and commercial insurance
plans. It can be delivered at clinics or via telemedicine. In India,
patients typically learn about this solution through a specialist’s
referral from private hospital chains, which do not accept insur-
ance. Therefore, the service is expected to be an out-of-pocket
expense.
• Revenue sharing with distribution channels. JOGO has signed
copromotion deals with large pharma and medical devices
companies that have complementary products in the neuromus-
cular diseases therapeutic area. As part of the deal, the channel
partners’ salesforces are leveraged to promote the DTx along
with their respective pharmaceutical compound or device.
• Franchise model. In India and Malaysia, JOGO has successfully
piloted a franchise concept whereby doctors refer patients to
a JOGO franchise, a dedicated physical clinic owned by the
franchisee. In this distribution model, JOGO receives a per-
centage of the revenues obtained from providing the serv ices to
patients, regardless of whether these serv ices are delivered in
the clinic or remotely.
As previously stated, too, there are geographical considerations for po-

tential DTx pricing models. We encourage DTx entrepreneurs to use
the checklist provided to determine the most suitable pricing model
for their solution while remaining compliant with all applicable local
and federal regulations.
Whichever approach DTx entrepreneurs choose to follow, poten-
tial channel partners w ill expect them to bring forth a motivated
pricing proposition. Channel partners can then look at their existing
portfolio and leverage their knowledge of the payer ecosystem to help
refine the offering in a way that works to their advantage too. DTx
innovators who have done their homework and have a solid business
plan, while being flexible about finding common ground, are the
ones that “make us sit up in our seats,” as one channel partner put it.

Therefore, it is incumbent on DTx players to get the right commercial
expertise and team in place that can help them define an optimal
pricing model.
Value-Based Care
Discussing DTx pricing models implies exploring value-based pricing
as well. It is a closely related concept and underlying philosophy for
many pricing models. Value-based pricing agreements pay providers
and DTx organizations based on the quality of care they offer, focus-
ing on improving patients’ clinical outcomes (quality) rather than the
volume of health care serv ices supplied or patients treated (quantity).
Although we do not attempt in this book to explain the intricacies of
value-based pricing in infinite detail, as health care becomes increas-
ingly outcome-driven, we do feel it is important for DTx innovators to
understand them in broad strokes. We raise the point as well b ecause
value-based pricing has evolved beyond the classical understanding
of value-based care, which shifted the focus from fee-for-service to health
outcomes over cost.
The contemporary definition of “value” is not just derived from mea
suring health outcomes against the cost of delivering the outcomes,
but it also considers f actors such as the appropriateness of care given
the social and individual context in which it is provided (figure 12.1).
Outcomes
= X
Appropriateness
Added value
for patients
Cost over the
full cycle of care
Figure 12.1. Defining value in health care from a patient’s perspective. Source:
Heuvel et al., Pathway to Success in Outcome-Based Contracting

10 Consolidated Markets United States Mature Markets
Health expenditure per capita (1,000 USD)
Germany
France Canada Australia
4 Japan United Kingdom
Spain
Italy
2
Future Markets Developing Markets
Mexico China
Brazil
1 2 3 4 5
Country experience with outcomes-based contracts
Figure 12.2. Country health expenditure (per capita) and experience with
outcomes-based contracts (on a scale of 1 to 5, with 1 being the least experienced
countries and 5 being the most experienced). Source: Heuvel et al., Pathway to Success in
Outcome-Based Contracting, 12 (citing original figure data sources as OECD, the World Bank,
Navigant, and KPMG Payer primary market research)
In line with this updated understanding of value-based care, pric-

ing models are starting to become more nuanced and granular as well.
T hese arrangements are actively being put in place (figure 12.2).
According to KPMG, in the United States between 2016 and 2019,
the therapeutic areas of cardiology, neurology, and endocrinology ac-
counted for more than 70% of the total outcomes-based deals between
the developers of health care solutions and providers and/or payers.5
Novartis, for example, has entered into such agreements with health
care providers by pledging to pay for any treatments that may still be
required beyond the 14-session standard of care for their Lucentis drug
that is used to treat age-related macular degeneration. Similarly, the
company’s pricing with insurers for its heart failure drug, Entresto, is
also based on outcomes, which are measured as the extent to which
the drug improves patients’ health. Novartis signed several contract-
ing models that tie pricing to patient outcomes, such as with health
insurers, like with Cigna and Aetna.

For their part, health plans are also becoming increasingly engaged
in the conversation around value-based pricing. In 2017, the nonprofit
health insurer Harvard Pilgrim Health Care in the United States agreed
with Amgen, the pharmaceutical company, to the first cardiovascular
outcomes-based refund contract. U nder the contract, Amgen would is-
sue a full rebate to the health plan if a patient continued to suffer
strokes or heart attacks after a full course of treatment with its anti-
body therapy Repatha.6
While this book encourages DTx innovators to take note of such
novel concepts around pricing, the industry is still fundamental in its
traditional and well-established payment mechanisms. As expressed
earlier, DTx teams must onboard the appropriate skills and experience
to match the channel partners’ and payers’ expectations and arrive
with business and commercial models in hand.
Conclusion
DTx pricing is the monetary reflection of its perceived (and demon-
strated) value in the eye of the beholder. While the FDA focuses on
such areas as DTx safety, insurers may be interested in how DTx de-
ployment may impact claims, and patients may be interested in how
the DTx solutions might help them live a healthier life for any associ-
ated out-of-pocket cost.
Earlier we emphasized the importance of demonstrating DTx safety
and effectiveness alongside robust health care economics outcomes,
which are collectively at the foundation of bridging innovation with
commercial success. Insufficiently accounting for all of these factors,
even with a powerful channel partner in place, may substantially im-
pede or even preclude commercializing DTx platforms in the compet-
itive health care landscape. Increasingly, evidence of effectiveness and
safety in the form of real-world evidence (RWE) may be required in
order for DTx platforms to be reimbursable. To generate sufficient
data and RWE, some DTx organizations are g oing “at-r isk”—being
paid only if a set of predefined health outcomes are being met—v ia

value-based and return-on-investment-based pricing models with pri-
vate and public channel partners.
An example of this approach was employed by Hello Heart, a DTx
focused on blood pressure management, to demonstrate that its plat-
form could successfully lower user blood pressure with outcomes-based
compensation instead of subscription-based payment by the client.7
Similar return-on-investment-based payment structures have also
been explored by Naluri for its chronic disease management solution.
T hese scenarios, while potentially risky for the solution provider, may
also play a catalytic role in scaling adoption due to the heavy empha-
sis placed on payment being driven by demonstrated efficacy. None-
theless, this reimbursement approach can be considered and is poten-
tially effective for the community of DTx developers. When considering
the broader landscape of how payers are viewing the f uture of DTx re-
imbursement, recent analysis in the insomnia space has shown that 9
out of 10 managed care organizations surveyed already have or w ill
have DTx reimbursement plans in place. To help payers better prepare
for a potential increase in DTx reimbursement activity, Express Scripts,
the largest pharmacy benefit-management organization in the United
States, has developed a digital health formulary to help payers system-
atically evaluate digital health solutions for potential coverage and
offerings to members.8
Regarding costs, we encourage DTx innovators to spend time truly
understanding the patient pathways to which their product applies,
the total direct and indirect cost of care in the current model (includ-
ing down to the unit and activity level, or patients’ potential out-of-
pocket expenses), and how incorporating a DTx solution could lead to
improved health outcomes at the same or lower cost (if any).
With value, cost, and outcomes data, DTx entrepreneurs can start ex-
ploring different pricing mechanisms that might exist in their respec-
tive geographies. If leveraging the elements and building blocks of the
technology industry, such as in a SaaS approach, for example, seems like
the fastest way to monetize DTx, we expect (and advise) DTx to build
first on traditional biopharma and medical technology companies’ busi-

ness and pricing models. New and hybrid models and reimbursement
pathways, combined with a variety of channels (i.e., B2C and B2B2C),
will emerge over time. As DTx are building evidence-based value propo-
sitions, our hope is that they not only transform the care delivered but
also advance the adoption of value-based contracting schemes.
Key Takeaways
• Value is in the eye of the beholder, so DTx innovators must seek to under-
stand potential channel partners’ and payers’ motivations for collaboration so
as to align their proposed pricing model.
• Given the focus and expected use of their solution, DTx innovators should
consider private-as well as public-sector reimbursement; this can be done in
sequence or tandem.
Notes
1. Organisation for Economic Co-operation and Development, Out-of-Pocket
Spending: Access to Care and Financial Protection.
2. Defining the needs prioritization for each potential cohort or customer
segment can be inspired by A. H. Maslow’s theory of human motivation and the five
categories of human needs that dictate an individual’s behavior (Maslow, “A Theory
of Human Motivation,” Psychological Review 50, no. 4 (1943), 370–96, https://doi.org
/10.1037/ h0054346).
3. Gleason-Comstock et al., “Willingness to Pay and Willingness to Accept in a
Patient-Centered Blood Pressure Control Study.”
4. Software as a serv ice, also known as “on-demand software,” is a software
licensing and delivery model in which software is licensed on a subscription basis
and delivered via the Internet.
5. Heuvel et al., Pathway to Success in Outcome-Based Contracting.
6. Heuvel et al., Pathway to Success in Outcome-Based Contracting.
7. Henze et al., “Moving Digital Health Forward: Lessons on Business Building.”
8. Prabhu and Menon, “Express Scripts a Pioneer: Launches Curated List of
Digital Therapeutics.”

• Conclusion
The Future of DTx
This chapter was written with the contribution of Dr. Eddie Martucci,
cofounder and CEO of Akili Interactive Labs in the United States.
In 2011, Eddie Martucci, cofounder of Akili Interactive Labs, was con-

vinced that technology was the best vehicle for providing novel treat-
ments for neurological and psychiatric diseases in ways drug-based
therapies alone were not able to do. Akili was born from what Eddie
perceived as a need and an aspiration for a new modality that would re-
imagine medicine and care delivery through software. T oday, the broad
modality of software addressing health conditions is a rapidly growing
field that provides patients with the opportunity to experience their
care in a more user-friendly, self-directed, even fun way, though no less
rigorously sound than traditional treatment pathways. Akili pursued
developing prescription DTx, where clinically validated therapeutic
software is prescribed by doctors to treat a disease. “When we shared
our vision back then, people would raise eyebrows and be quite skeptical
of something like a video game interface that could be as effective as
traditional medicine prescribed by a doctor,” remembers Eddie.
249

It took more than a decade for the biopharmaceutical and medical
device industry, regulators, and early DTx pioneers like Akili to grow
the industry and address key “firsts” for DTx, such as product devel-
opment and validation, conducting robust clinical trials with DTx soft-
ware to demonstrate clinical evidence on par with traditional thera-
peutics, and identifying a pathway for obtaining market authorization
by regulatory bodies. T hose efforts have paved the way for a wave of
DTx solutions that seemed like science fiction less than a decade ago
but are actually being realized in the hands of patients t oday.
The vision and the slowly growing momentum behind DTx were el-
evated to near the top of the health care agenda when the COVID-19
pandemic began to wreak havoc and reveal the weaknesses of the
established care models. T hose models, based on physical check-
ins, brick-a nd-mortar waiting rooms, and in-person physician con-
sultations on the one hand, and bedeviled by challenges around
treatment adherence and patient monitoring on the other, were sud-
denly primed for disruption. The COVID-19 crisis gave rise to new
partnership opportunities and alternative solutions as patients be-
gan to cancel traditional doctor’s appointments and health organ
izations began to offer options to access care differently—in patients’
homes and communities instead of at hospitals and clinics. Many
physicians and patients who may have never had a video conference
call before became comfortable with telemedicine-based serv ices and
remote monitoring platforms and applications.
Similarly, and just as pervasively, nearly every family experienced
some version of the stress, anxiety, and psychological issues previously
relegated to diagnosed mental health conditions. Thus, COVID-19 si
multaneously had a major direct impact on growing the awareness of
and need for new, scalable, and accessible interventions. Enter digital
options for treatment and care that, perhaps even more so than tele-
medicine, w ill be h
ere to stay for the long term.
Digital health tools, including DTx, pharma companion apps, and
wellness platforms, reached an inflection point in popularity, leading

to a gold rush of experimentation by health care stakeholders. The reg-
ulatory environment adapted seemingly instantaneously, and many
of the barriers that had previously held back progress in operational-
izing digital health technology, such as availability, cost, and market
adoption, seemed to shift overnight. In addition to drugs and durable
medical devices, consumers began asking for digital treatments—and
physicians are now prescribing DTx. Thanks in part to the urgency cre-
ated by COVID-19, and cruising on the wings of broad digital trans-
formation, “software as medicine” became a reality.
DTx are now poised to reimagine health care, and the future is ex-
citing. Software-based interventions are interacting with human
physiology in ways never seen before, such as incorporating music to
assist with walking in p eople with neurological disorders, treating
Alzheimer’s using noninvasive light-based stimulation techniques,
or losing oneself in a fun video game that is actually rewiring neural
processing and treating attention dysfunction in c hildren. If break-
throughs such as t hese are any indication, DTx are g oing to revolu-
tionize health care and eventually be integrated into mainstream
medicine. But this revolution is more than just the dawn of new
medical products; w e’re seeing the emergence of an entirely new pil-
lar of medicine that can engage and respond to patients in dramati-
cally new ways, putting the role of the patient more central than ever
before.
The Role of Patients in a New Era of Medicine

Patients are ready for change. They are more informed and empow-
ered than ever before, expecting a high-end user experience in all as-
pects of their lives, and being seen not as conditions to treat but as
individuals to heal. With DTx, patients like you and I, and future gener-
ations, will experience medicine differently. Many of these new technol-
ogy benefits will be unobstructive to the patient and, in many cases, ac-
tually exciting (Akili, for example, has a design ethos that their treatment
Conclusion 251

products are entertaining enough that, in the moment of the DTx ex-
perience, the patient might forget they are “taking medicine” at all).
Indeed, DTx offer inspiration and empowerment to patients in new
ways because the experiences they provide can be intentionally de-
signed to be engaging and exciting. For instance, using virtual reality
to ease pain and video game interfaces to improve ADHD impair-
ments can be fun and captivating. Even more exciting, the patient is
finally empowered to play a central role in the management of their
condition by directing the management’s evolution in response to
digitally delivered insights, reminders, and prompts, or “nudges” in
the simplest case. This empowerment is further supported by real-
time product adaptations and highly personalized product features
designed to allow personalized benefit.
Furthermore, the real-world data and feedback of DTx users can be
of g reat value to DTx organizations, physicians, and care teams, as they
can inform improvements to the interaction design, underlying soft-
ware, and eventually disease treatment provided by these tools. In a
positive turn on the “uses of data” controversy we’ve seen with social
media platforms, companies are using data from DTx to actually en-
hance their products’ core value proposition to patients over time, al-
lowing an evolution of safety, effectiveness, and patient experience
not only in future products but in the exact same products that are in
patients’ hands. The future of medicine can be rapidly adaptive, and
the incentive of business and user benefit w ill be aligned.
A vast population of patients—more than many health care lead-
ers and providers expect—are ready to experience new treatment mo-
dalities.1 It is they who w ill own the design and evolution of the DTx
ecosystem, and it is they who w ill eventually determine the demand
for DTx. As Eddie recalls, “One of the very first days that our first prod-
uct was available on the market, without any product marketing, a
physician called Akili to know more about our treatment because one
of her patients had already asked her twice for a prescription.” Patients
are activated and anticipating the new wave.

A New Patient-Physician Relationship
Although DTx are not designed to replace traditional medicine or
drugs, they offer a new type of treatment modality and a new prom-
ise for patients and physicians that can be used as part of an overall
treatment program, either alone or alongside traditional medicine. But
do DTx threaten to replace physicians?
This question has sparked a series of long debates, but the conclu-
sion is clearly no.2 We see technology, and within that DTx, as quite
the opposite: enablers of a better patient-physician relationship, which
in itself is a key driver of clinical outcomes.3 If technology is designed
properly, it should enhance this relationship by involving other mem-
bers of the clinic care team (e.g., technicians, nurses, or social work-
ers), as well as the patients’ caregivers, whose critical role is often over-
looked. And technology can offer rich data to drive meaningful
dialogue between patients and their physicians.
Focusing on treatment for c hildren with ADHD, Akili developed
ADHD Insight, a behavior-tracking app and online dashboard that
helps parents monitor changes in their child’s ADHD symptoms.
Through the data and feedback collected from the c hildren and their
care teams, parents can understand how their child is d oing and feel-
ing on a daily basis. Eddie says, “We expected the first piece of parent
feedback to be about the nuances of what the data means about spe-
cific symptoms, but the number one t hing we hear from parents is that
it’s actually helping them have better conversations with their children,
as well as more meaningful conversations with the physicians and
their staff.” This type of unexpected insights should not be overlooked:
families care deeply about their relationships, perhaps more so than
their specific symptoms—and digital platforms can enhance this.
Today in traditional medicine t here are few data to guide such
physician-patient conversations beyond the patient’s memory and rec-
ollection of how they feel. In mental health, in particular, having
more precise real-world data is especially relevant as it can mean the
difference between continuing with the same approach to treatment,
Conclusion 253

modifying it, or switching to an entirely different one. DTx can col-
lect such data e very day, or potentially many times per day. And when
it’s shared with physicians, patients, and the broader care team, it cre-
ates the foundation for a new type of relationship in medicine that
doesn’t exist today.
Evolving Business Models

The digital therapeutics landscape will evolve drastically over the next
few years, with the anticipated arrival of new applications and in-
creased reimbursement coverage by commercial health plans and
government payers. In the case of specific disease treatments, DTx w ill
continue to be prescription-based products, but our hope is that the
prescribing ability of physicians w ill grow relative to the limitations
they currently face in many countries in terms of who can prescribe
what digital tool for what purpose. DTx formularies w ill ultimately ex-
pand and be less restricted not only because of different safety pro-
files compared to traditional medicine, but also b ecause with time
health care practitioners w ill gain a more comprehensive understand-
ing of how t hese products work. Independent of formularies, as more
medical practitioners get their first and second experiences prescrib-
ing DTx, the “new modality” barrier w ill recede and doctors w ill be-
come advocates for those treatments that most help their patients, just
like any other medicine.
Although prescribed DTx w ill likely be a powerful pillar in the ther-
apeutic market with many products, that d oesn’t mean that there w ill
be a single business model around their use or integration with other
modalities. Pharma’s evolving relationship with DTx serves as a good
example. Over time, some pharmaceutical companies w ill strengthen
their DTx-related capabilities with dedicated resources or departments,
while o thers w
ill remain focused on their traditional core business,
thereby becoming potential distribution partners. For their part,
DTx organizations are starting to realize that the pharmaceutical in-
dustry can be a natural ally in the development, distribution, and

commercialization of DTx, and many commercialization options,
alone or in a partnership, w ill be v iable.
T here is, therefore, no one-size-fits-a ll business model for DTx.
Akili, for example, uses a market-based approach. In the United States,
the company is building its own commercial distribution model that
is a hybrid of key tactics from the pharmaceutical and consumer soft-
ware industries, whereas in Japan it has signed a partnership with a
pharmaceutical company that has deep expertise in local regulatory,
marketing, and sales mechanics. Other DTx organizations may use
this as inspiration . . . or not. Time w ill tell.
One t hing is for sure: the amount of real-world data and evidence
generated and collected by DTx tools far exceeds that obtained in the
course of traditional drug-based therapies. This opening of the flood-
gates on data collection w ill necessarily lead to new business models,
such as value-based agreements and other outcomes-based arrange-
ments, providing greater incentives for treatments that work and less
for t hose that do not. For now, the priority for most DTx innovators,
entrepreneurs, and companies (besides obtaining regulatory clearance)
is how to provide patients access to DTx. It’s the responsibility of both
DTx organizations as well as the larger ecosystem, including public and
private payers, to crack the tough nut of moving from experimental
programmatic spending to claims-based reimbursement, and, quite
simply, to paying for products that are clinically proven and can en-
hance patients’ lives, irrespective of their modality.
From Hype to Reality

here is a lot of hype in the digital health space, which is commonly
T
accompanied by exaggerated expectations that might lead to disillu-
sionment. DTx are not a miracle digital alternative designed to replace
traditional medicine. Neither are they a marketing tactic that purports
to convince you to leave your doctor and surrender into technology’s
arms. To rule out any temptation to conflate DTx discussions with
marketing speak, we need to collectively realize the potential of t hese
Conclusion 255

digital health tools beyond the hype. For that, DTx organizations need
to invest in engaging physicians and patients early on, building trust
with them over time, and having proactive communication with all
stakeholders at all stages of DTx development, all the while focusing
on the most critical cornerstone of generating solid clinical evidence.
Why is clinical evidence so important? As more and more startups
and organizations jump on the business of health care bandwagon, the
number of products, platforms, and applications that make outland-
ish claims and are being pushed onto physicians and patients, in-
creases. To stand out from this crowd, DTx innovators need to arm
themselves with the best tool that separates hype from reality: clini-
cal evidence. It is the key aspect of DTx that makes all the difference
and that should be the focal point of any conversation with physicians
around a DTx trial, pilot, or implementation.
Technology is only a part of a solution and it cannot reinvent health
care on its own. For this reason, our hope is that the experiences and
insights contained in this book will serve as a how-to guide in your DTx
journey and help you overcome the challenges that certainly lie ahead
as you move closer to implementing your solution into existing mod-
els of care. It’s time to realize the potential of DTx and improve pa-
tient outcomes at scale. When patients need it, and you have proof it can
help, it’s worth the hard work to make it happen . . . and all signs are point-
ing to that happening in a very big way.
Notes
1. Mobiquity Inc., “COVID-19: Ensuring a Quality Patient Experience with the
Rise of Digitisation in a Healthcare Setting”; Sham, Accenture 2019 Digital Health
Consumer Survey.
2. Ahuja, “The Impact of Artificial Intelligence in Medicine on the Future Role
of the Physician”; Goldhahn et al., “Could Artificial Intelligence Make Doctors
Obsolete?”
3. Gordon, Phillips, and Beresin, “The Doctor–Patient Relationship.”

Acknowle dgments
We are deeply grateful to Chong Yap Seng, Lien Ying Chow Professor in
Medicine and Dean of the Yong Loo Lin School of Medicine, National
University of Singapore (NUS Medicine). Under his visionary leadership,
WisDM was established to redefine the concept of patient impact through
practice-changing, multidisciplinary health care innovation. We thank
Professor Chong for supporting this book project.
This book is based on the collective experience of our many contribu-
tors and experts who share our belief in the importance of raising aware-
ness about digital therapeutics (DTx), a new class of medicine, but also
about what it takes to develop, validate, and implement a new technol-
ogy, especially in health care settings. We are sincerely grateful for their
contributions and support. Our teams at the Institute for Digital Medi-
cine (WisDM) at NUS Medicine, N.1 Institute for Health, and Depart-
ment of Biomedical Engineering in the College of Design and Engineer-
ing at NUS have immensely contributed to the book by sharing their
expertise. We would like to thank Associate Professor Christopher L.
Asplund, Associate Professor B. T. Thomas Yeo, Associate Professor Ja-
son Kai Wei Lee, and their team members for the privilege of serving as
their collaborators. We would like to thank, too, Gavin Teo for helpful
discussions and for his leadership and dedication toward advancing en-
trepreneurship and innovation in DTx.
The book is also the result of a community of patients and highly tal-
ented and committed researchers, engineers, and clinicians from the
faculties and departments of NUS, NUS Medicine, and the National Uni-
versity Hospital in Singapore. They tirelessly push the boundaries in
health care to improve patients’ lives. We are fortunate to reside within
one of the few ecosystems in the world that seamlessly integrates e very
domain that is essential to realizing health care impact at scale and the
257

full potential of digital health—from technology and trial design to be-
havioral sciences and policy.
We would like to thank our amazing copyeditor, Gergana Koleva, who
helped ready the manuscript, as well as everyone who reviewed the many
drafts of the manuscript and generously gave us their time toward this
project. We are deeply appreciative of their insights.
To Poonam Rai, Khew Si Ying, the NUS Office of Legal Affairs, and
NUS Development Office, thank you for being an amazing part of our
team and community. Your continuous support has been a catalyst
toward everything from the successful launches of so many projects to
the establishment of the Patient Impact Fund and the completion of this
book.
To Sally Toh, Gladys Sim, and Michael Lim from the NUS Medicine
communications team; Dr. Raja Kamal and Frank Lavin from the NUS
Medicine International Council; Ron Kaufman; and the NUS Medicine
finance and contracts teams. Thank you for always serving as the gate-
way for conveying our innovations to the broader community. Your ef-
forts have played a vital role in bridging our work with our surrounding
ecosystem. We are deeply appreciative of all of your unwavering support
from the very beginning of this book project, through its development
process, and to its launch.
This book would not have been possible without the enthusiasm of
our agents, Nick Wallwork and Chris Newson, and our wonderful edi-
tors, Joe Rusko, Robert Brown, Alena Jones, and Adelene Jane Medrano.
We are grateful for their comments and editorial feedback.
Last but not least, we would like to thank our families and close ones
for their unwavering support while we embarked on this exciting and
ambitious project.
258 Acknowledgments

Appendix A
AI and Its Potential Use in DTx:
A Brief Overview of Frequently Asked Questions
Alexandria Marie Remus, PhD
Alexandria Marie Remus, PhD, is a DTx development specialist at the N.1

Institute for Health at the National University of Singapore.
What Is Artificial Intelligence?

To understand the potential of applying artificial intelligence (AI) t oward
digital therapeutics (DTx), one first must understand the basic concepts of
AI. AI is highly embedded in the world around us; however, misalignment
between its technical definition and daily use of the term has been confus-
ing for many aiming to enter the field. At its core, AI is a branch of com-
puter science defined as the simulation of h uman intelligence in machines,
computers, or robots. In this sense, its application centers on programming
machines to think and act like humans (or sometimes animals) by “teach-
ing” them to learn, recognize, understand, respond, decide, and solve
problems on their own.
AI is being applied to many different industries, including finance,
education, transportation, health care, retail, logistics, and sports. To
understand the true potential application of AI, however, one must also be
aware of the different types and subsets of the field. A general schematic
representation of the complexity of AI according to its “strength” and the
function for which it is being deployed is depicted in figure A.1. In this
common understanding, the internal hierarchy of this field is determined
along two main branches, or types: capability (Type 1) and functionality
(Type 2). T hese two branches are further broken down into subcategories.
Type 1: Capability of AI
Type 1 classifies AI according to its capabilities and encompasses three
categories: artificial narrow intelligence (ANI), artificial general intelli-
gence (AGI), and artificial super intelligence (ASI). ANI, also referred to as
weak AI, describes AI that powers a machine to perform a particu lar task
but does not enable the machine to have decision-making capacity, and the
259

Weak Al
(Artificial Narrow Intelligence)
Type 1: Strong Al
Capability (Artificial General Intelligence)
Super Al
(Artificial Super Intelligence)
Artificial Intelligence
Reactive Al
Limited Memory Al
Type 2:
Functionality
Theory of Mind Al
Self-Aware Al
Figure A.1. Hierarchy of AI. Source: Adapted by permission from Springer Nature:
Zelinka et al., “Artificial Intelligence in Astrophysics,” 3
machine w ill never “pass” as a h

uman in any other task or function outside
this predefined capacity. AGI, also referred to as strong AI, describes AI that
programs machines with the ability to learn and apply their intelligence to
solve any problem and performs on par with h umans. ASI, also known as
super AI, describes AI in which machines are programmed to evolve and
eventually become more intelligent than humans. T hese three categories of
AI are described in more detail below.
Weak AI
AI technologies that are commercialized or used in research t oday largely
fall within the category of weak AI. Weak AI is widely leveraged in commer-
cial applications, science, business, and health care. Its most common use is
in programming machines to recognize commands and return searchable
information or to detect patterns in data and classify new data accordingly.
Virtual assistants such as Apple’s Siri, Amazon’s Alexa, and Microsoft’s
Cortana are good examples of this. T hese voice-activated devices “listen”
for what they are “taught” to recognize, classify those cues, and then
produce a more or less accurate response. For instance, when you ask Alexa
260 Appendix A

to “play Rage Against the Machine on Spotify,” the voice assistant is
programmed to understand key words and phrases such as “play,” “rage
against the machine,” and “Spotify,” and w ill respond by playing Rage
Against the Machine over the speaker. Chatbots that are often available on
a business or a hospital website that are used to direct customers’ inquiries
run on the same principle, although their input format may be not voice but
text instead. Other applications of weak AI use computers to process big
data and convert them into practical bites of information by detecting
patterns and making predictions; use cases include medical diagnoses,
personalized song recommendations, purchase recommendations, and even
email spam filters. In summary, weak AI machines do what they are told to
do but do not “comprehend” the meaning of what they are asked to do or
the results they produce and cannot perform outside of predefined tasks.
While the majority of these applications are deemed as positive and
beneficial, since they enhance our experiences as citizens, consumers, and
patients, weak AI does not come without risks. For example, systems that
incorporate weak AI may fail unintentionally. Examples include disrup-
tions in power supply or chaotic driving by autonomous vehicles. A weak AI
system can also be programmed to do something good or beneficial but
respond with a destructive method for achieving that goal. An example of
this can be a self-driving car that is programmed to send a passenger home
as quickly as possible but accomplishes this task by speeding through red
lights. B
ecause of these inherent risks of current AI architectures, integrat-
ing AI capabilities to perform any new task should be meticulously
planned, tested, and evaluated.
Strong AI
The next evolutionary stage of AI is the category of strong AI. Strong AI
could endow machines with human-level intelligence but currently only
exists in theory. This category of AI describes the hypothetical intelligence
of a computer program that has the capacity to undertake any intellectual
task a h uman can, such as reasoning, solving puzzles, making judgements,
planning, learning, and communicating, as well as possessing conscious-
ness, objective thoughts, self-awareness, and sentience. In contrast to weak
AI, strong AI does not follow specific cues or commands to respond but
instead processes the data with grouping and association rules to generate
an independent and unpredictable response. What this means in practice is
that t here isn’t a set of predefined responses the machine can choose from
and reply with, which is essentially what Alexa and its peers do. Instead, it
AI and Its Potential Use in DTx 261

produces a novel or unpredictable result, similar to how we might expect a
human to respond when asked a question.
While AGI is currently hypothetical, a recent survey of AI experts predicted
its emergence by 2060.1 Although this date is debated, with the rapid develop-
ment of AI, it is reasonable to believe that applications of strong AI might
appear within our lifetime or by century’s end. Developments behind GPT (the
“engine” of ChatGPT and Bing) are examples of significant steps toward AGI.
Super AI
The third category of AI is super AI. This is the kind of AI that most people
think of when they talk about robots taking over the world—a common
plotline in science fiction movies and books. In t hese fictional scenarios, AI
surpasses h uman intelligence and ability. Again, super AI is purely specula-
tive at this point and w ill most likely remain an element of science fiction
for the foreseeable f uture.
Type 2: Functionality of AI
Type 2 classifies AI according to its functionalities and has four categories:
reactive AI, limited memory AI, theory of mind AI, and self-aware AI.
While this grouping is independent from the grouping based on capabili-
ties of AI, the functionalities have similar underlying logic and applica-
tions. T
hese four categories of AI are described in more detail below.
Reactive AI
Reactive AI, as the name suggests, has the functionality to “react” to a given
situation and context by doing exactly what it has been programmed to do.
This type of AI cannot form memories or use past experiences to make
decisions. As a result, it does not learn and evolve and is therefore the most
basic type of AI. The most well-known example of reactive AI is the super-
computer Deep Blue created by IBM that was programmed to play chess and
challenge Garry Kasparov, the world chess champion, in 1996 and 1997.2
L imited Memory AI
The functionality of limited memory AI is what the term implies—a
machine that has the capability to retain some information that it has
learned or observed and to use it in conjunction with preprogrammed data.
With this functionality, machines can observe the environment, detect
patterns or changes, and respond accordingly. They do so by applying
previous training on data from big datasets that are stored in the machine’s
262 Appendix A

memory as a reference model. T hese models can improve over time, unlike
the static reactive AI, b
ecause they incorporate machine-learning and
deep-learning models, described later in this chapter. Limited memory AI
is used in most AI applications today and common examples include image
classification, autonomous cars, chatbots, and virtual assistants.
Theory of Mind AI
Theory of mind AI describes a theoretical AI functionality whereby
machines have decision-making capabilities that are equivalent to h uman
decision-making. With this functionality, machines would be able to un-
derstand and remember not just situations but also emotions and adjust
their behavior just as h
umans do in diff erent social interactions and
contexts. While researchers and scientists are making tremendous leaps
in developing theory of mind AI, we are still some time away from seeing
it in practice. T
here are, however, two well-known sociable robots, Kis-
met, developed by Dr. Cynthia Breazeal from Massachusetts Institute of
Technology in the late 1990s, and Sophia, developed by Hanson Robotics
in 2016, that have been developed using elements of theory of mind AI.
Self-Aware AI
Like theory of mind AI, self-aware AI currently only exists in theory and is
imagined as an extension of theory of mind AI. Self-aware AI machines
would not only be able to recognize and replicate h uman actions but also
think for themselves, form their own desires, and understand their own
feelings. This would be the most complex type of AI and—like super AI—
at present exists only in the realm of science fiction. It is unknown if and
when h umans w ill be capable of developing machines with self-aware AI
functionalities.
Subsets of AI
Now that we have a brief and general understanding of the two types of AI
and their subcategories, let’s have a look at the different subsets of AI in order
to complete our picture of what AI is and how it can be applied, particularly
with respect to DTx. These subsets of AI are loosely categorized by the type of
analytical method or algorithm used to achieve their respective AI capability
and functionality. Figure A.2 illustrates how the subsets of AI relate to each
other and what applications they encompass. Machine learning (ML), and its
own subset, deep learning (DL), are the most common applications of AI
within a health care context and have the potential to be leveraged into DTx.

Artificial Intelligence Machine Learning Deep Learning
• Computer vision • Supervised learning • Ability to reason
• Natural language processing • Unsupervised learning • Physical enablement
• Voice recognition • Reinforcement learning
• Robotics & motion
• Planning & optimization
• Knowledge capture
Figure A.2. Different subsets of AI. Source: Adapted from Zaleha H et al., “Intelligent
Locking System Using Deep Learning for Autonomous Vehicle in Internet of T
hings,” 568
Machine Learning
Within the overarching universe of AI, machine learning is its first and
most common subset. Its foundational purpose is to “teach” computers
how to learn without the necessity to be programmed for specific tasks. To
achieve this, the machines are fed data and equipped with an algorithm
that enables them to learn from the data and then use the insights to make
predictions or other intelligent decisions. In ML models, the data input can
be anything from numbers, to words, to images, to sounds, to even clicks.
Most applications of AI t oday use machine learning in some shape or form.
Delving further, t here are three approaches of machine learning, based
on the expected input and output of the underlying algorithm: supervised
learning, unsupervised learning, and reinforcement learning. An AI model
using supervised learning relies on previously labeled data with the aim of
calculating and predicting an outcome. By comparison, an AI model using
unsupervised learning is trained on unlabeled data and without any
guidance with the aim of discovering underlying patterns. Lastly, an AI
model using reinforcement learning works by interacting with its environ-
ment, with no predefined labeled or unlabeled data, with the aim of
inferring on its own a series of actions and, as the name suggests, reinforc-
ing that learning through f uture iterations. The choice of one or another of
these ML types depends on the problem to be solved.
Deep Learning
Deep learning is one of the most rapidly developing subsets of machine
learning and AI t oday. It describes a machine’s capability to mimic the
workings of the h uman brain and perform human-l ike tasks without
human input. This subset of machine learning is based on a neural network
264 Appendix A

model, which is a network of algorithms that attempt to replicate the
perception and thought process of the h
uman brain. A simplified neural
network has three main components: (1) an input layer, where data enter
the network; (2) one or more hidden, or computational, layers, where at
least one algorithm processes the input data by applying learnable par
ameters; and (3) an output layer, where the conclusions of the model, which
might take the form of predictions or diagnoses, emerge. More complex
neural network models, known as deep neural networks, are comprised of
many layers that interact to analyze the data and shape the final output.
Generally, deep learning is the leading AI approach used in autonomous
vehicles and speech recognition.
What Are Examples of Current Applications of AI in Health Care?

Current uses of AI in health care settings range from supporting personnel
with day-to-day tasks, optimizing administrative workflow, and improving
image analysis for diagnostics to streamlining remote patient monitoring.
A few of these applications and some selected examples are detailed below.
Augmenting Diagnostics
One of the most widely used AI applications in clinical settings is diagnos-
tics, where they are deployed to sift through electronic health record data or
medical scans and discard instances in which t here are no signs of disease,
as well as provide further detail in cases where t here may be a basis for
concern. In d oing so, such applications both save time for busy clinicians
and reduce h uman error that may occur by inadvertently overlooking
problematic cases. In other scenarios, ML and DL algorithms are leveraged
to analyze patient clinical data—such as vital signs, symptoms, images, and
other test results—w ith the aim of predicting probable diagnoses and
outcomes. This is made possible by having previously trained the algo-
rithms to analyze patterns in large patient datasets and to predict suscepti-
bility to developing specific diseases based on t hose patterns, some of which
may be too subtle for clinicians to recognize. Another application of AI in
the context of diagnostics involves training algorithms to recognize
anomalies—such as tumors in biopsies and various imaging modalities. In
fact, a recent review identified t hese AI technologies to have similar or
better accuracy than trained clinicians in detecting disease from medical
images.3 Some examples of AI platforms for augmenting diagnostics include
PathAI (https://w ww.pathai.com), Buoy Health (https://w ww.buoyhealth
.com), and BEHOLD.AI (https://behold.ai).

Augmenting Drug Discovery and Development
The traditional research and development process requires thousands of
human hours and billions of dollars to develop a drug candidate, test it in
clinical trials, and potentially receive a regulatory approval. Yet only a
small percentage of new drugs are successfully validated and brought to
market. To compress the time it takes to go through these R&D stages, AI
is increasingly being used to screen large compound libraries, predict which
potential compounds are likely to be effective for a certain condition, and
direct a small subset of selected compounds for subsequent experimental
testing. This approach results in saved time and money as compared to
experimentally testing the w hole compound library. Another common AI
application in the context of drug development is automating labor-
intensive and repetitive tasks, such as image analysis, which also leads to a
reduction in human hours spent. Yet another application is using AI
algorithms to screen large drug libraries to identify drugs that may be
safely repurposed for treating additional indications than t hose for which
they w ere originally developed. In fact, many labs globally harnessed this
AI approach during the emergence of the COVID-19 pandemic to identify
possible therapies while vaccine development was being pursued in
parallel. Some examples of existing AI applications for augmenting drug
discovery and development include IDentif.AI,4 BioXcel Therapeutics
(https://w ww.bioxceltherapeutics.com), Engine Biosciences (https://
enginebio.com), and BenevolentAI (https://w ww.benevolent.com).
What Is the Potential for AI in DTx?

As discussed throughout the book, DTx platforms are on a trajectory to
transform patient care. As a reminder, the Digital Therapeutics Alliance
defines DTx platforms as technologies that deliver evidence-based,
clinically validated therapeutic interventions, driven by validated software
programs, to prevent, manage, or treat medical conditions. DTx clinical
trials and their eventual deployment can generate massive volumes of data
in real time at unprecedented accuracy and frequency—a perfect match for
AI technologies. While we only highlighted a few current applications of AI
in health care, we hope it is evident that AI has the potential to take DTx
and patient care to the next level.
One of the most promising applications of the u nion of AI with DTx is
the potential to move health care away from one-size-fits-a ll approaches
to delivering more individualized, patient-centric interventions. DTx offer
266 Appendix A

the ability to remotely collect a large amount of user data over a number
of time points. T
hese large datasets can provide deep insights into longitu-
dinal trends and acute changes for a user, which are opportunistic for ML
and DL models. Such algorithms can then be trained and harnessed to
deliver personalized interventional care for users. An application of a
personalized preventative intervention may involve a DTx that promotes
behavioral changes that can decrease a patient’s risk for developing a
disease, such as the transition from a prehypertensive to hypertensive
stage by promoting an increase in physical activity. Such a DTx may
monitor a prehypertensive person’s activity levels and use ML algorithms
to identify periods of low activity to then send personalized reminders to
encourage physical activity or reward the user when they perform physical
activity. The resulting behavioral change of increased activity may prevent
this individual from progression t oward more severe stages of hyperten-
sion and may even remove them from the prehypertensive stage.
In a similar DTx that does not harness AI, all users may be subject to the
same preventative intervention at the same time points regardless of their
own activity levels. While a number of users may still benefit from such
DTx intervention, the treatment may be suboptimal, impact compliance,
and may not benefit the majority of users. Multiple other examples can be
found throughout the book. Overall, the incorporation of AI in DTx solutions
has the potential to deliver more personalized care to large populations
without the need for a large number of specialized health care providers
required to deliver such care, which in turn may improve outcomes.
What Are Some Data and Regulatory Challenges Posed

by AI in DTx?
While the potential for AI-driven DTx solutions seems undeniable, t here
are also challenges to overcome and risks that must be assessed during the
course of development as well as implementation. For example, most often
AI algorithms require large amounts of data to be trained on. This requires
access to mass user data in early development before an AI-driven DTx may
be proven effective and beneficial. This mass collection and storage of
patient data also requires AI-driven DTx solutions to implement stringent
privacy and security measures to prevent data leaks. Developers also need
to be aware of bias incorporated into the implemented AI algorithm if it is
not trained on a diverse patient sample and must consider the safety of
deploying such algorithms in a diverse sample.

Another challenge is that the AI-driven DTx technologies must still
undergo the regulatory approval process. Unlike traditional therapeutics
that involve the intake of drugs, the substantial amounts of data collected
by an AI-enabled DTx make it possible for the underlying algorithms to
improve and evolve over time. Thus, AI-DTx version 1.0 can become AI-DTx
version 1.1, which can then become AI-DTx version 1.2, and so on. This
potential for technological evolution raises the question of what happens
if AI-DTx 1.0 gains regulatory approval but is soon a fter improved and
upgraded—w ill the approval remain valid for AI-DTx 1.1 and subsequent
software updates? As this field expands, the momentum for developing
frameworks and guidelines for DTx that incorporate AI is continuously
growing.
Signaling its readiness for this new phase, in January 2021 the US Food
and Drug Administration released a plan for establishing a regulatory
approach for evaluating and improving evolving AI and ML algorithms as
well as advancing real-world performance pilots to provide clarity on what
a real-world evidence program would look like for AI-and ML-based
software as a medical device (SaMD). Similarly, in October 2021 Singa-
pore’s Ministry of Health (MOH), Health & Sciences Authority, and
Integrated Health Information Systems codeveloped the MOH Artificial
Intelligence in Healthcare Guidelines for both developers and implemen-
tors of AI tools in health care based in Singapore.
This is an important step in the right direction. Additional contribu-
tions to this arena from the community of international regulatory bodies
w ill serve as an important collaborative foundation, as AI technology is
also evolving during this period. Additionally, the development and
adoption of established regulatory frameworks may be among the primary
catalysts that rapidly advance the field.
What Is the F uture of AI in DTx?

DTx have a yet-to-be-f ully-tapped potential to deliver evidence-based
treatment interventions to patients across many conditions in a uniquely
efficient way. Its combination with AI offers insights and interventional
opportunities that traditional and other digitalized therapies do not. As
the field continues to grow and specific regulatory frameworks emerge, it
is expected that AI-powered DTx w ill transform the conceptualization and
delivery of health care. T hese opportunities to improve patient care and
outcomes are endless and exciting.
268 Appendix A

Notes
1. Dilmegani, “When Will Singularity Happen? 995 Experts’ Opinions on AGI.”
2. Goodrich, “How IBM’s Deep Blue Beat World Champion Chess Player Garry
Kasparov.”
3. Liu et al., “A Comparison of Deep Learning Performance against Health-Care
Professionals in Detecting Diseases from Medical Imaging.”
4. Blasiak et al., “IDentif.AI: Rapidly Optimizing Combination Therapy Design
against Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) with
Digital Drug Development.”

Appendix B
How to Translate Evidence-Based Theories to the Design of
Digitally Delivered M
ental Health Interventions
This appendix was coauthored with Dr. Geck Hong Yeo, research fellow
at the N.1 Institute for Health (N.1), and Matt Oon, CEO and founder of
Acceset, a Singapore-based social enterprise. Acceset aims to transform
the management of m ental health care through an anonymous text-based
therapy involving trained volunteers overseen by licensed therapists and
psychiatrists. The teams at WisDM and N.1 have been working with Acceset
to develop prospective protocols to validate, through qualitative and quantitative
research, the use of its technology to provide emotional support to Singaporean
youths struggling with anxiety and/or depression.
In our experience working with startups involved in digital health (DH)

development, we have found that many tend to build their solutions based
on their own experiences and applied frameworks. The theoretical resources
tend to be overlooked or viewed as not being directly applicable enough to
the solution being developed and/or the condition being addressed.
Importantly, evidence-based theoretical frameworks (EBTFs) can guide
decisions on which variables to investigate and which features to include in
the solution.
Consider the story of Siam (a pseudonym), a patient battling depression.
There is a constant inner voice that amplifies a sense of anxiety or a sense of
worthlessness and calibrating this requires a very intentional and intensive
process with the therapist, whom I’m already seeing once a week for two hours.
It is not always easy to get a consultation slot and these sessions are not the
most affordable . . . but you know that you need them to learn to dissociate
from thoughts that prevent one from feeling good about oneself—to deal with
the shame, fear and constant judgment. . . . Sometimes you feel lonely and you
have a lot of negative thoughts and feelings that eat away your worth as a
person. How one views the world, one’s outlook of life, is often a mirror image
of their internal state of mind. I can see myself in the mirror and believe that
270

I am not worthy. Then this can influence the outlook I have on life in general
and events that occur in my life. Your f amily, friends, and most p eople in the
society d on’t understand what you are going through, nor respect the fact that
you need your space to heal. This has made recovery a truly difficult process, as
I cannot control how o thers view me and how supportive they would be to the
things I do for my own mental wellness. —Siam, 32 years old
Siam’s story is particularly insightful. It underlines the difficulties he, his

family, and his doctor face in managing his disease, as well as the chal-
lenges we face as a society in aiding people suffering from depression. Like
those suffering from any m ental illness, they continue to endure signifi-
cant stigma even t oday, including general social distancing stigma (being
perceived as dangerous and unpredictable, as someone best avoided) and
being seen as having a personal weakness, or being “weak not sick.”1
The way society views mental health needs to change. For Siam and many
others, the priority is to recognize that being depressed does not imply being
“weak.” Doing so will assist and empower him to better comprehend his
condition and realize that he might need help. This is precisely where DTx
technology can be successfully leveraged to facilitate access to mental health
literacy, resources, and support in a user-centric and anonymous format.
DTx may also offer the possibility to engage persons suffering from depres-
sion beyond therapist consultations. For example, it can help Siam manage
his symptoms with personalized cognitive and physical exercises as well as
monitor his progress. If DTx platforms have the potential to help individuals
with a m ental health condition, they should be designed in a way that can
effect meaningful change to these individuals, including long-term behavior
modification. To do so, DTx needs to rest on solid scaffolding that bridges
theoretical and clinically driven underpinnings.
Interventions can be based on varying segments along the theoretical to
clinical/user continuum. Where applicable, a theoretical basis for interven-
tion design—such as the exploration of mattering-based digital peer
support—could indeed be important when designing m ental health
solutions. This foundation provides a set of principles guiding how different
aspects or elements of social phenomena relate to each other and can direct
our use of DTx. A theory is defined as a set of principles governing how
different aspects or elements of social phenomena relate to each other.2 As
an organizing framework, it accounts for what is known, explains the
current state, and predicts f uture phenomena. T hese principles of a theory
are generated from scientific methods and grounded in empirical evidence.
How to Translate Evidence-Based Theories 271

To emphasize the importance of evidence in supporting eventual DTx
deployment, the development and evaluation of DTx platforms informed
by principles that are generated from scientific methods and grounded in
empirical evidence may ultimately be more clinically effective than t hose
that a ren’t. For t hese reasons, we recommend that researchers, practitioners,
and budding DTx entrepreneurs build teams with experience in validating
evidence-based theories and frameworks for the purpose of developing
high-quality DTx solutions. H ere Dr. Geck Hong Yeo and Matt Oon share
some of their experiences translating theoretical and empirical research into
real-world applications of designing DTx for addressing m ental health needs.
Why Is An Evidence-B ased Theoretical Framework Useful

for DTx in General?
The Story behind Acceset, by Mr. Oon
I have been a caregiver to my m other, who has schizophrenia, since I was
7 years old. The responsibilities I had to take on throughout the years seeded
my interest for m ental health issues. In the very beginning, being her
caregiver inspired me to volunteer for initiatives supporting the m ental
well-being of young people in Singapore. This is where I discovered my passion
for innovation and driving social change. I also witnessed that physical clinics
and peer networks had limitations in engendering help-seeking behavior
among youth. They d idn’t provide sufficient protection of their identity, nor
did they give them control over the medium and mode of emotional disclosure.
The development of Acceset stemmed from the belief that lending a listening
ear to struggling youths in Singapore was what they r eally needed.
Today Acceset provides anonymous text-based therapy through a digital
letter-writing platform, which offers a safe space for t hose who want to share
their concerns anonymously and receive support. We strive to make emo-
tional support easily accessible to youths while educating them on mental
health care, by leveraging technology as well as qualitative and quantitative
research.
Working on Theories with Acceset, by Dr. Yeo

Specific to Acceset, a sound evidence-based framework at the conception
stage would address critical questions, such as why Acceset is focusing on
online peer support. What form of peer support should Acceset enable?
What features of the online peer support system should Acceset include
that w
ill facilitate peer disclosure online?
272 Appendix B

It is true for Acceset, as it is for all digital health technologies, that it is
critical to identify a sound, relevant, and empirically supported theoretical
framework. The alternative and too common “post hoc” approach, whereby
developers would design and implement the DH solution and then try to
retrofit it with a relevant theoretical framework, leads to downstream
issues, such as failing to include relevant or key features of DH that are
needed to address the health concerns that the DH was originally intended
for (or including irrelevant ones). It also presents potential cost-
effectiveness concerns associated with the technology needed for the DH.
The next step, therefore, is to identify an appropriate EBTF.
How Do You Determine If an Evidence-B ased Theoretical

Framework Is Relevant to Your Mental Health Solution?
From our experience, it is useful to consider the following five questions
to help you identify a sound and relevant EBTF for your m ental health
solution:
1. Does the theory describe and explain the links between the outcomes
that the DTx is designed to measure and intervene upon?
2. Is t here information outlining the application of the theory to a
mental health intervention, especially in the envisioned intervention
context (e.g., features of the individuals and community, modes of
intervention)?
3. Has this theory already been applied in longitudinal studies (i.e.,
investigating change in mental health behaviors over time) and/or
translational research (i.e., examining the mechanism of change
between the intervention and mental health outcomes/changes, or
the steps or path in which the intervention leads to mental health
outcomes/changes)?
4. Is t here evidence for the application of the theory to an intervention
that has demonstrated change in the expected magnitude and
direction in m ental health outcomes or behaviors?
5. Is t here information/evidence that clearly defines the mechanism
of action?
Mattering as an Evidence-B ased Theory for Acceset, by Mr. Oon

For Acceset, it is the mattering theory that informed our solution. The
mattering theory refers to the extent to which we make a difference to
the world in which we live.3 It resonates strongly with youths’ lived

experiences of anxiety and depression. For this group of individuals, it
can instill a sense of significance and offers a specific positive framing for
addressing m ental health issues. We are therefore focused on conveying
an increased sense of mattering through digital interventions that can
boost the well-being of the adolescents.
Deciding on Mattering as a Potential Framework for Acceset, by Dr. Yeo

We went through the five key questions in thinking about and deciding on
a theory that would be the most appropriate in the design of Acceset as a
digital peer intervention in support of adolescents’ mental well-being.
Here is the outline of the thought processes for three of t hese questions:
1. Does the theory describe and explain the links between the outcomes that the
DTx is designed to measure and intervene upon?
Mattering is comprised of attention, importance, and reliance. Recent
studies seek to understand the role of mattering in adolescents’ well-being,
such as being a potential “active ingredient” in interventions aimed at
mitigating adolescent anxiety and depression.4 Thus, we are investigating
mattering as a relevant EBTF for developing and strengthening Acceset’s
solution.
Acceset’s platform underscores the building of interpersonal relation-
ships that enhance their sense of mattering. Through the disclosure of
emotionally laden experiences online with one’s peers, and their peers’
supportive responses, adolescents receive attention that is much needed
for helping them manage how they express and regulate their emotions.
This interpersonal exchange may improve their sense of importance and
reliance, as adolescents feel that their needs and concerns are heard, and
they can rely for support on others who share similar experiences. This
may also contribute to a sense of meaning and purpose and overall
well-being.
2. Is there information outlining the application of the theory to a mental health
intervention, especially in the envisioned intervention context?
Theory and research on mattering has been applied to understand
and support adolescents’ anxiety and depression, including their lived
experiences comprising irrational thoughts, fear, lack of control, and
feelings of insignificance.5 Research on mattering demonstrated signifi-
cant gender differences—specifically, females perceived a higher overall
sense of mattering and domain-specific mattering related to f amily,
school, and friends—w ith only one study documenting nongendered
274 Appendix B

effects.6 Such information is pertinent in the design of a solution as a
mental health intervention.
In terms of age differences in mattering, t here are also findings that
hint at the relative importance of mattering to the f amily and peer
domains for young adults and adolescents, respectively.7
3. Has this theory already been applied in longitudinal studies and/or translational
research?
Research documented the steps in how mattering impacts psychological
well-being. Mattering is related to one’s selfhood and is distinguished from
other important self domains, such as self-esteem and self-presentation.
Increasing one’s sense of mattering, in turn, increases self-esteem, which
contributes to or supports youths’ well-being.8
Conclusion
By translating theoretical and empirical research into the conceptualization,
design, operationalization, and real-world application of DTx, DTx entrepre-
neurs can efficiently and effectively leverage technology to improve the m ental
health of young people and of society as a w hole. We therefore recommend
exploring the role of evidence-based theoretical frameworks as a guide to help
pinpoint features to incorporate in a given solution as well as clinical outcomes
to assess by way of providing evidence for the effectiveness of that solution.
Therefore, in the design stage, we encourage you to think about t hese next
questions as key pointers in applying an EBTF in the design of your solution:
1. What does the evidence-based theory specify to be an effective
component or “active ingredient” of the intervention?
2. What are the key characteristics or features of the solution that
function as the intervention for the gap you are addressing?
3. How do t hese defining characteristics of the solution map onto the
effective components or active ingredients specified by the theory?
4. What are the underlying operating principles, assumptions, and
relations between the effective components or active ingredients
outlined by the evidence-based theory?
5. How does this evidence-based theory explain how different effective
components or active ingredients of your solution interact with one
another in producing an effect to function as an intervention?
The application of an evidence-based theoretical framework should not be
l imited to the conceptualization of your solution but also considered for its
operationalization, as we discuss in Part III of the book.

Notes
1. Subramaniam et al., “Stigma t owards P eople with Mental Disorders and Its
Components.”
2. Moller et al., “Applying and Advancing Behavior Change Theories and
Techniques in the Context of a Digital Health Revolution.”
3. Elliott, Kao, and Grant, “Mattering: Empirical Validation of a Social-
Psychological Concept.”
4. Marshall and Tilton-Weaver, “Adolescents’ Perceived Mattering to Parents
and Friends”; Flett et al., “Academic Buoyancy and Mattering as Resilience Factors
in Chinese Adolescents.”
5. Flett, Burdo, and Nepon, “Mattering, Insecure Attachment, Rumination, and
Self-Criticism in Distress among University Students”; Matera, Bosco, and Merin-
golo, “Perceived Mattering to F amily and Friends, Self-Esteem, and Well-Being.”
6. Marshall and Tilton-Weaver, “Adolescents’ Perceived Mattering to Parents
and Friends”; Elliott, Colangelo, and Gelles, “Mattering and Suicide Ideation”;
Marshall, “Do I M atter?”; Dixon, Scheidegger, and McWhirter, “The Adolescent
Mattering Experience.”
7. Dixon, Scheidegger, and McWhirter, “The Adolescent Mattering Experience”;
Finkel and Baumeister, Advanced Social Psychology: The State of the Science.
8. Flett and Nepon, “Mattering versus Self-Esteem in University Students:
Associations with Regulatory Focus, Social Feedback, and Psychological Distress.”
276 Appendix B

Bibliography
Adler, Nancy E., and Ann E. K. Page, eds. Cancer Care for the Whole Patient:
Meeting Psychosocial Health Needs. Washington, DC: National Academies
Press, 2008. https://w ww.ncbi.nlm.nih.gov/ books/NBK4015.
Ahmad, Amar S., Nick Ormiston-Smith, and Peter D. Sasieni. “Trends in
the Lifetime Risk of Developing Cancer in G reat Britain: Comparison of
Risk for those Born from 1930 to 1960.” British Journal of Cancer 112,
no. 5 (2015): 943–47. https://doi.org/10.1038/ bjc.2014.606.
Ahuja, Abhimanyu S. “The Impact of Artificial Intelligence in Medicine on
the F uture Role of the Physician.” PeerJ 7 (2019): e7702. https://doi.org
/10.7717/peerj.7702.
Akili Interactive. “Akili and Shionogi Announce Strategic Partnership to
Develop and Commercialize Digital Therapeutics in Key Asian Markets.”
Press release, March 17, 2019. https://w ww.a kiliinteractive.com/news
-collection/a kili-and-shionogi-announce-strategic-partnership-to
-develop-and-commercialize-digital-therapeutics-in-key-asian-markets.
Akili Interactive. “Akili Announces FDA Clearance of EndeavorRx™ for
Children with ADHD, the First Prescription Treatment Delivered
through a Video Game.” Press release, June 15, 2020. https://w ww
.a kiliinteractive.com/news-collection/a kili-announces-endeavortm
-attention-treatment-is-now-available-for-children-w ith-attention
-deficit-hyperactivity-disorder-adhd-a l3pw.
Alagizy, H. A., M. R. Soltan, S. S. Soliman, N. N. Hegazy, and S. F. Gohar.
“Anxiety, Depression, and Perceived Stress among Breast Cancer
Patients: Single Institute Experience.” M iddle East Current Psychia-
try 27 (2020): 29. https://doi.org/10.1186/s43045- 020- 00036-x.
American Diabetes Association. “The Cost of Diabetes.” March 22, 2018.
https://w ww.diabetes.org /resources/statistics/cost-diabetes.
American Diabetes Association. “Insulin Myths and Facts.” Clinical Diabetes
25, no. 1 (2007): 39–40. https://doi.org/10.2337/diaclin.25.1.39.
277

American Psychological Association. “What Is Cognitive Behavioral
Therapy?” PTSD Clinical Practice Guideline. Accessed November 17,
2022. https://w ww.apa.org/ptsd-g uideline/patients-and-families
/cognitive-behavioral.pdf.
Anderson, Annick, Deborah Borfitz, and Kenneth Getz. “Global Public
Attitudes about Clinical Research and Patient Experiences with Clinical
Trials.” JAMA Network Open 1, no. 6 (2018): e182969. https://doi.org/10
.1001/jamanetworkopen.2018.2969.
Angelis, Aris, David Tordrup, and Panos Kanavos. “Socio-Economic Burden
of Rare Diseases: A Systematic Review of Cost of Illness Evidence.” Health
Policy 119, no. 7 (2015): 964–79. https://doi.org/10.1016/j.healthpol.2014
.12.016.
Anthony, Scott D., Mark W. Johnson, Joseph V. Sinfield, and Elizabeth J.
Altman. The Innovator’s Guide to Growth: Putting Disruptive Innovation to
Work. Boston: Harvard Business Review Press, 2008.
Aref-Adib, Golnar, Tayla McCloud, Jamie Ross, Puffin O’Hanlon, Victoria
Appleton, Sarah Rowe, Elizabeth Murray, Sonia Johnson, and Fiona
Lobban. “Factors Affecting Implementation of Digital Health Interven-
tions for People with Psychosis or Bipolar Disorder, and their F amily
and Friends: A Systematic Review.” Lancet Psychiatry 6, no. 3 (2019):
257–66. https://doi.org/10.1016/S2215-0366(18)30302-X .
Arnhold, Madlen, Mandy Quade, and Wilhelm Kirch, “Mobile Applications
for Diabetics: A Systematic Review and Expert-Based Usability Evalua-
tion Considering the Special Requirements of Diabetes Patients Aged
50 Years or Older.” Journal of Medical Internet Research 16, no. 4 (2014):
e104. https://doi.org /10.2196/JMIR .2968.
Barclay, Laurie. “Exubera Approved Despite Initial Lung Function Con-
cerns.” Medscape Medical News, February 9, 2006. https://w ww.medscape
.com/v iewarticle/523294.
Barker, Pierre M., Amy Reid, and Marie W. Schall. “A Framework for
Scaling Up Health Interventions: Lessons from Large-Scale Improve-
ment Initiatives in Africa.” Implementation Science 11, no. 1 (2015): 1–11.
https://doi.org/10.1186/s13012-016-0374-x.
Beede, Emma, Elizabeth Baylor, Fred Hersch, Anna Iurchenko, Lauren
Wilcox, Paisan Ruamviboonsuk, and Laura M. Vardoulakis. “A Human-
Centered Evaluation of a Deep Learning System Deployed in Clinics for
the Detection of Diabetic Retinopathy.” In Proceedings of the 2020 CHI
Conference on Human Factors in Computing Systems, (New York: Association
278 Bibliography

for Computing Machinery, 2020), 1–12. https://doi.org/10.1145/3313831
.3376718.
Belluck, Pam, and Rebecca Robbins. “FDA Approves Alzheimer’s Drug
Despite Fierce Debate Over W hether It Works.” New York Times, June 7,
2021. https://w ww.nytimes.com/2021/06/07/ health/aduhelm-fda
-a lzheimers-drug.html.
Bestsennyy, Oleg, Greg Gilbert, Alex Harris, and Jennifer Rost. “Tele-
health: A Quarter-Trillion-Dollar Post-COVID-19 Reality?” McKinsey &
Company, July 9, 2021. https://w ww.mckinsey.com/industries/ healthcare
-systems-and-services/our-insights/telehealth-a-quarter-trillion-dollar
-post-covid-19-reality.
Blasiak, Agata, Jhin Jieh Lim, Shirley Gek Kheng Seah, Theodore Kee,
Alexandria Remus, De Hoe Chye, Pui San Wong, et al. “IDentif.AI:
Rapidly Optimizing Combination Therapy Design against Severe Acute
Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) with Digital Drug
Development.” Bioengineering & Translational Medicine 6, no. 1 (2021):
e10196. https://doi.org/10.1002/ btm2.10196.
Blue Note Therapeutics. “Cancer-Related Distress.” Accessed July 4, 2021.
https://bluenotetherapeutics.com/why.
Boersma, Peter, Lindsey I. Black, and Brian W. Ward. “Prevalence of Multiple
Chronic Conditions among US Adults, 2018.” Preventing Chronic Dis-
ease 17 (2020). https://doi.org//10.5888/pcd17.200130.
Bombard, Yvonne, G. Ross Baker, Elaina Orlando, Carol Fancott, Pooja
Bhatia, Selina Casalino, Kanecy Onate, Jean-L ouis Denis, and Marie-
Pascale Pomey. “Engaging Patients to Improve Quality of Care: A System-
atic Review.” Implementation Science 13 (2018): 98. https://doi.org/10
.1186/s13012-018-0784-z.
Braithwaite, Jeffrey, Peter D. Hibbert, Adam Jaffe, Les White, Christo-
pher T. Cowell, Mark F. Harris, William B. Runciman, et al. “Quality of
Health Care for Children in Australia, 2012–2013.” JAMA 319, no. 11
(2018): 1113–24. https://doi.org/10.1001/jama.2018.0162.
Bridge Point Capital. “Advancement of Digital Therapeutics Leads to a
Huge Market with Enormous Potential.” Accessed October 15, 2020.
https://w ww.bridgepoint.capital/post/advancement-of-digital
-therapeutics-leads-to-a-huge-market-w ith-enormous-potential.
Brown, Marie T., and Jennifer K. Bussell. “Medication Adherence: WHO
Cares?” Mayo Clinic Proceedings 86, no. 4, (2011): 304–14. https://doi.org
/10.4065/mcp.2010.0575.
Bibliography 279

Buttorff, Christine, Teague Ruder, and Melissa Bauman. Multiple Chronic
Conditions in the United States. Vol. 10. Santa Monica, CA: Rand, 2017.
https://w ww.rand.org/pubs/tools/ TL221.html.
Byrne, Jane. “DIA 2021 Set to Shed Some Light on Decentralized T rials
Process, Accelerated Approval Pathways.” BioPharma-Reporter, June 16,
2021. https://w ww.biopharma-reporter.com/A rticle/2021/06/16/DIA
-2021-set-to-shed-some-light-on-decentralized-trials-process-accelerated
-approval-pathways.
Califf, Robert M., Deborah A. Zarin, Judith M. Kramer, Rachel E. Sherman,
Laura H. Aberle, and Asba Tasneem. “Characteristics of Clinical T rials
Registered in ClinicalTrials.gov, 2007–2010.” JAMA 307, no. 17 (2012):
1838–47. https://doi.org/10.1001/jama.2012.3424.
CB Insights. “The Top 12 Reasons Startups Fail.” August 3, 2021. https://
www.cbinsights.com/research/startup-failure-reasons-top.
Centers for Diseases Control and Prevention. “Health and Economic Costs
of Chronic Diseases.” Accessed November 17, 2022. https://w ww.cdc.gov
/chronicdisease/about/costs.
Centers for Disease Control and Prevention. National Diabetes Statistics
Report 2020: Estimates of Diabetes and Its Burden in the United States.
Atlanta, GA: Centers for Disease Control and Prevention, US Depart-
ment of Health and Human Serv ices, 2020. https://w ww.cdc.gov
/diabetes/pdfs/data /statistics/national-diabetes-statistics-report.pdf.
Centers for Disease Control and Prevention. “Workplace Health Promotion:
Depression Evaluation Measures.” Accessed November 17, 2022. https://
www.cdc.gov/workplacehealthpromotion/ health-strategies/depression
/evaluation-measures/index.html.
Centro Criptológico Nacional [National Cryptologic Center]. “Requisitos de
seguridad para aplicaciones de cibersalud [Security Requirements for
eHealth Applications].” September 7, 2020. https://w ww.ccn-cert.cni.es
/comunicacion-eventos/comunicados-ccn-cert/10471-requisitos-de
-seguridad-para-aplicaciones-de-cibersalud.html.
Chan, Eugene Y., and Najam U. Saqib. “Privacy Concerns Can Explain
Unwillingness to Download and Use Contact Tracing Apps when
COVID-19 Concerns Are High.” Computers in Human Behavior 119
(June 2021): 106718. https://doi.org/10.1016/j.chb.2021.106718.
Chen, Sharon, and Sterling Wong. “Singapore Beats Hong Kong in Health
Efficiency: Southeast Asia.” Bloomberg, September 17, 2014. https://w ww
.bloomberg.com/news/articles/2014-09-18/singapore-beats-hong-kong
-in-health-efficiency-southeast-asia.
280 Bibliography

Christensen, Clayton M., Taddy Hall, Karen Dillon, and David S. Duncan.
“Know Your Customers’ ‘Jobs to be Done.’ ” Harvard Business Review 94,
no. 9 (2016): 54–62. https://hbr.org/2016/09/ know-your-customers
-jobs-to-be-done.
Choi, Mi Joo, Hana Kim, Hyun-Wook Nah, and Dong-W ha Kang, “Digital
Therapeutics: Emerging New Therapy for Neurologic Deficits a fter
Stroke.” Journal of Stroke 21, no. 3 (2019): 242–58. https://doi.org/10
.5853/jos.2019.01963.
Chudasama, Rajesh K., A. M. Kadri, Apurva Ratnu, Mahima Jain, and
Chandrakant P. Kamariya. “Magnitude of Gestational Diabetes Melli-
tus, Its Influencing Factors, and Diagnostic Accuracy of Capillary Blood
Testing for Its Detection at a Tertiary Care Centre, Rajkot, Gujarat.” In-
dian Journal of Community Medicine 44, no. 2 (2019): 142–46. https://
pubmed.ncbi.nlm.nih.gov/31333293.
Clement, Innocent, Andreas Lorenz, Bernhard Ulm, Anne Plidschun, and
Stephan Huber. “Implementing Systematically Collected User Feedback
to Increase User Retention in a Mobile App for Self-Management of Low
Back Pain: Retrospective Cohort Study.” JMIR mHealth and uHealth 6,
no. 6 (2018): e10422. https://doi.org/10.2196/10422.
Clinical Trials Transformation Initiative (CTTI). CTTI Recommendations:
Decentralized Clinical T rials. Durham, NC: CTTI, September 2018.
https://ctti-clinicaltrials.org/w p-content/uploads/2021/06/CTTI_ DCT
_Recs.pdf.
Clinical Trials Transformation Initiative (CTTI). Developing Novel Endpoints
Generated by Digital Health Technology for Use in Clinical T rials. Durham,
NC: CTTI, April 2022. https://ctti-clinicaltrials.org/w p-content/uploads
/2022/03/CTTI-Digital-Health-Trials-Novel-Endpoint-Acceptance
-Recommendations.pdf.
Clinical Trials Transformation Initiative (CTTI). Digital Health Trials:
Recommendations for Delivering an Investigational Product. Durham, NC:
CTTI, July 8, 2021. https://ctti-clinicaltrials.org/w p-content/uploads
/2021/07/CTTI _Delivering _ an_Investigational_Product_ Recommen
dations.pdf.
Coravos, Andrea, Jennifer C. Goldsack, Daniel R. Karlin, Camille Nebeker,
Eric Perakslis, Noah Zimmerman, and M. Kelley Erb. “Digital Medicine:
A Primer on Measurement.” Digital Biomarkers 3, no. 2 (2019): 31–71.
https://doi.org /10.1159/000500413.
Cosgrove, Lisa, Ioana Alina Cristea, Allen F. Shaughnessy, Barbara Mintzes,
and Florian Naudet. “Digital Aripiprazole or Digital Evergreening?
Bibliography 281

A Systematic Review of the Evidence and Its Dissemination in the
Scientific Literature and in the Media.” BMJ Evidence-Based Medicine 24,
no. 6 (2019): 231–38. https://doi.org/10.1136/ bmjebm-2019-111204.
Darden, Michael, Colin A. Espie, Jenna R. Carl, Alasdair L. Henry, Jenni-
fer C. Kanady, Andrew D. Krystal, Christopher B. Miller. “Cost-
Effectiveness of Digital Cognitive Behavioral Therapy (Sleepio) for
Insomnia: A Markov Simulation Model in the United States.” Sleep 44,
no. 4 (2021): zsaa223. https://doi.org/10.1093/sleep/zsaa223.
David, Daniel, Ioana Cristea, and Stefan G. Hofmann. “Why Cognitive
Behavioral Therapy Is the Current Gold Standard of Psychotherapy.” Fron-
tiers in Psychiatry 9 (2018): 4. https://doi.org/10.3389/fpsyt.2018.00004.
Davidai, Shai, Thomas Gilovich, and Lee D. Ross. “The Meaning of Default
Options for Potential Organ Donors.” Proceedings of the National Acad
emy of Sciences 109, no. 38 (2012): 15201–5. https://doi.org/10.1073/pnas
.1211695109.
Davis, Christian, Emily Burgen, and Guoqing J. Chen. “Out-of-Pocket Costs
for Patients with Type 2 Diabetes Mellitus.” Diabetes 67, supplement 1
(2018): 2324-PUB. https://doi.org /10.2337/db18-2324-PUB.
Dean Foods and Livongo Health. Livongo Drives 1.4x ROI in Year 1 for Dean
Foods: Enrollment and Activation Best Practices Accelerate Outcomes (case
study). Dallas, TX: Dean Foods and Livongo Health, 2021. https://
content.livongo.com/DEANFOODS/CaseStudy.pdf.
Derksen, Frans, Jozien Bensing, and Antoine Lagro-Janssen. “Effective-
ness of Empathy in General Practice: A Systematic Review.” British
Journal of General Practice 63, no. 606 (2013): e76-e84. https://doi.org/10
.3399/ bjgp13X660814.
Dilmegani, Cem. “When Will Singularity Happen? 995 Experts’ Opinions
on AGI.” AI Multiple, August 3, 2017. https://research.aimultiple.com
/artificial-general-intelligence-singularity-timing.
Digital Therapeutics Alliance. Digital Health Industry Categorization.
Arlington, VA: Digital Therapeutics Alliance, November 2019. https://
dtxalliance.org/w p-content/uploads/2019/11/DTA_ Digital-Industry
-Categorization_ Nov19.pdf.
Digital Therapeutics Alliance. Digital Therapeutics Definition and Core Princi
ples. Arlington, VA: Digital Therapeutics Alliance, November 2019.
https://dtxalliance.org/w p-content/uploads/2019/11/DTA_ DTx
-Definition-and- Core-Principles.pdf.
Digital Therapeutics Alliance. “DTx by Country.” Accessed April 5, 2022.
https://dtxalliance.org/understanding-dtx/dtx-by-country.
282 Bibliography

Digital Therapeutics Alliance. DTx Product Best Practices. Arlington, VA:
Digital Therapeutics Alliance, November 2019. https://dtxalliance.org
/w p-content/uploads/2019/11/DTA_ DTx-Product-Best-Practices_11.11
.19.pdf.
Digital Therapeutics Alliance. “DTx Product Profile: BlueStar.” Accessed
November 17, 2022. https://dtxalliance.org/productbluestar.
Digital Therapeutics Alliance. Ensuring Appropriate Quality, Access, and
Utilization of Digital Therapeutics. Arlington, VA: Digital Therapeutics
Alliance, April 2020. https://dtxalliance.org/w p-content/uploads/2020
/04/DTx _Quality_ Access_Utilization _Worksheet.pdf.
Digital Therapeutics Alliance. “Finding Value in Digital Therapeutics
during COVID-19.” March 19, 2020. https://dtxalliance.org/2020/03/19
/finding-value-in-dtx-products-during-covid-19.
Digital Therapeutics Alliance. “Understanding DTx.” May 21, 2021.
https://dtxalliance.org/understanding-dtx.
Dixon, Andrea L., Corey Scheidegger, J. Jeffries McWhirter. “The Adoles-
cent Mattering Experience: Gender Variations in Perceived Mattering,
Anxiety, and Depression.” Journal of Counseling & Development 87, no. 3
(2011): 302–10. https://doi.org/10.1002/j.1556- 6678.2009.tb00111.x.
Egede, Leonard E., Mulugeta Gebregziabher, Clara E. Dismuke, Cheryl P.
Lynch, R. Neal Axon, Yumin Zhao, and Patrick D. Mauldin. “Medication
Nonadherence in Diabetes: Longitudinal Effects on Costs and Potential
Cost Savings from Improvement.” Diabetes Care 35, no. 12 (2012):
2533–39. https://doi.org/10.2337/dc12-0572.
eHealth Stakeholder Group. Proposed Guiding Principles for Reimbursement of
Digital Health Products and Solutions. Luxembourg: Medtech Europe.
Accessed April 5, 2022. https://w ww.medtecheurope.org/w p-content
/uploads/2019/04/30042019_eHSGSubGroupReimbursement.pdf.
Elliott, Gregory, Melissa F. Colangelo, and Richard J. Gelles. “Mattering
and Suicide Ideation: Establishing and Elaborating a Relationship.”
Social Psychology Quarterly 68, no. 3 (2005): 223–38. https://doi.org/10
.1177/019027250506800303.
Elliott, Gregory, Suzanne Kao, and Ann-Marie Grant. “Mattering: Empiri-
cal Validation of a Social-Psychological Concept.” Self and Identity 3,
no. 4 (2010): 339–54. https://doi.org/10.1080/13576500444000119.
Espie, Colin A., Simon D. Kyle, Chris Williams, Jason C. Ong, Neil J.
Douglas, Peter Hames, and June S. L. Brown. “A Randomized, Placebo-
Controlled Trial of Online Cognitive Behavioral Therapy for Chronic
Insomnia Disorder Delivered via an Automated Media-R ich Web
Bibliography 283

Application.” Sleep 35, no. 6 (2012): 769–81. https://doi.org/10.5665
/sleep.1872.
European Commission. “Internal Market, Industry, Entrepreneurship, and
SMEs: Conformity Assessment.” Accessed November 1, 2021. https://ec
.europa.eu/growth/single-market/goods/ building-blocks/conformity
-assessment_en.
The European Union Medical Device Regulation, May 26, 2021. https://
eumdr.com.
Farr, Christina. “Proteus Digital Health, Once Valued at $1.5 Billion, Files
for Chapter 11 Bankruptcy.” CNBC, June 16, 2020. https://w ww.cnbc
.com/2020/06/15/proteus-digital-health-once-worth-1point5-billion
-fi les-for-chapter-11.html.
Fiagbe, J., S. Bosoka, J. Opong, W. Takramah, W. K. Axame, R. Owusu, P. A.
Parbey, M. Adjuik, E. Tarkang, and M. Kweku. “Prevalence of Controlled
and Uncontrolled Diabetes Mellitus and Associated Factors of Con-
trolled Diabetes among Diabetic Adults in the Hohoe Municipality of
Ghana.” Diabetes Management 7, no. 5 (2017): 343–54.
Federal Ministry of Health. “Driving the Digital Transformation of
Germany’s Healthcare System for the Good of Patients: The Act to
Improve Healthcare Provision through Digitalisation and Innovation
(Digital Healthcare Act—DVG).” Last updated April 22, 2020. https://
www.bundesgesundheitsministerium.de/digital-healthcare-act.html.
Finkel, Eli J., and Roy F. Baumeister (eds.). Advanced Social Psychology: The
State of the Science. Oxford: Oxford University Press, 2019.
Fitzsimmons-Craft, Ellen E., C. Barr Taylor, Andrea K. Graham, Shiri
Sadeh-Sharvit, Katherine N. Balantekin, Dawn M. Eichen, Grace E.
Monterubio, et al. “Effectiveness of a Digital Cognitive Behavior Therapy–
Guided Self-Help Intervention for Eating Disorders in College Women: A
Cluster Randomized Clinical Trial.” JAMA Network Open 3, no. 8 (2020):
e2015633. https://doi.org/10.1001/jamanetworkopen.2020.15633.
Flett, Gordon L., Ron Burdo, and Taryn Nepon. “Mattering, Insecure
Attachment, Rumination, and Self-Criticism in Distress among Univer-
sity Students.” International Journal of M ental Health and Addiction 19
(2020): 1300–13.
Flett, Gordon L., and Taryn Nepon. “Mattering versus Self-Esteem in
University Students: Associations with Regulatory Focus, Social
Feedback, and Psychological Distress.” Journal of Psychoeducational
Assessment 38, no. 6 (2020): 663–74. https://doi.org/10.1177
/0734282919890786.
284 Bibliography

Flett, Gordon L., Chang Su, Liang Ma, and Lianrong Guo. “Academic Buoy-
ancy and Mattering as Resilience Factors in Chinese Adolescents: An
Analysis of Shame, Social Anxiety, and Psychological Distress.” Interna-
tional Journal of Child and Adolescent Resilience 2, no. 1 (2014): 37–45.
https://w ww.ijcar-rirea.ca/index.php/ijcar-rirea/article/view/159.
Forrester. “The Biggest Digital Health Deal of All Time: Teladoc and Livongo.”
Forbes, August 7, 2020. https://w ww.forbes.com/sites/forrester/2020/08
/07/the-biggest-digital-health-deal-of-all-time-teladoc-and-livongo.
Fraunhofer Institute for Algorithms and Scientific Computing SCAI.
“Preventive Treatment for Alzheimer’s and Parkinson’s Disease.”
Fraunhofer-Gesellschaft, June 1, 2018. https://w ww.fraunhofer.de/en
/press/research-news/2018/june/preventive-treatment-for-a lzheimers
-and-parkinsons-disease.html.
Gallagher, Paul. “GP Appointments Should Be at least 15 Minutes in
Future, Leading Doctors Say.” iNews, May 21, 2019. https://inews.co.u k
/news/ health/g p-appointments-15-m inutes-standard-length-royal
-college-293235.
Garavand, Ali, Mohammah Mohseni, Heshmatollah Asadi, Manal Etemadi,
Mohammad Moradi-Joo, and Ahmad Moosavi. “Factors Influencing the
Adoption of Health Information Technologies: A Systematic Review.” Elec-
tronic Physician 8, no. 8 (2016): 2713–18. https://doi.org/10.19082/2713.
Gleason-Comstock, Julie, Alicia Streater, Allen Goodman, James Janisse,
Aaron Brody, LynnMarie Mango, Rachelle Dawood, and Phillip Levy.
“Willingness to Pay and Willingness to Accept in a Patient-Centered
Blood Pressure Control Study.” BMC Health Services Research 17, no. 1
(2017): 1–6. https://doi.org /10.1186/s12913-017-2451-5.
Granja, Conceição, Wouter Janssen, and Monika Alise Johansen. “Factors
Determining the Success and Failure of eHealth Interventions: System-
atic Review of the Literature.” Journal of Medical Internet Research 20,
no. 5 (2018): e10235. https://doi.org/10.2196/10235.
Greene, Jessica, Judith H. Hibbard, Rebecca Sacks, Valerie Overton, and
Carmen D. Parrotta. “When Patient Activation Levels Change, Health
Outcomes and Costs Change, Too.” Health Affairs 34, no. 3 (2015):
431–37. https://doi.org/10.1377/ hlthaff.2014.0452.
Greenhalgh, Trisha, and Chrysanthi Papoutsi. “Spreading and Scaling Up
Innovation and Improvement.” BMJ 365 (May 2019). https://doi.org/10
.1136/ bmj.l2068.
Greenhalgh, Trisha, Joseph Wherton, Chrysanthi Papoutsi, Jennifer
Lynch, Gemma Hughes, Susan Hinder, Nick Fahy, Rob Procter, and Sara
Bibliography 285

Shaw. “Beyond Adoption: A New Framework for Theorizing and Evaluat-
ing Nonadoption, Abandonment, and Challenges to the Scale-Up,
Spread, and Sustainability of Health and Care Technologies.” Journal of
Medical Internet Research 19, no. 11 (2017): e367. https://doi.org/10.2196
/jmir.8775.
Grefkes, Christian, and Gereon R. Fink. “Recovery from Stroke: Current
Concepts and Future Perspectives.” Neurological Research and Practice 2
(2020): 17. https://doi.org/10.1186/s42466-020-00060- 6.
Goldhahn, Jörg, Vanessa Rampton, Branco Weiss, and Giatgen A. Spinas.
“Could Artificial Intelligence Make Doctors Obsolete?” BMJ 363 (2018).
https://doi.org /10.1136/ bmj.k4563.
Gonzalez, Jeffrey S., Lauren A. McCarl, Deborah J. Wexler, Enrico Cagliero,
Linda Delahanty, Tiffany D. Soper, Valerie Goldman, Robert Knauz, and
Steven A. Safren. “Cognitive–Behavioral Therapy for Adherence and
Depression (CBT-AD) in Type 2 Diabetes.” Journal of Cognitive Psychother-
apy 24, no. 4 (2010): 329–43. https://doi.org/10.1891/0889-8391.24.4.329.
Goodrich, Joanna. “How IBM’s Deep Blue Beat World Champion Chess
Player Garry Kasparov.” IEEE Spectrum, January 25, 2021. https://
spectrum.ieee.org / how-ibms-deep-blue-beat-world-champion-chess
-player-garry-k asparov.
Gordon, Christopher, Margot Phillips, Eugene V. Beresin. “The Doctor–
Patient Relationship.” In Massachusetts General Hospital Handbook of
General Hospital Psychiatry, 6th ed., edited by Theodore A. Stern,
Gregory L. Fricchione, Ned H. Cassem, Michael Jellinek, and Jerrold F.
Rosenbaum, 15–23. Philadelphia: W. B. Saunders, 2010. https://doi.org
/10.1016/B978-1-4377-1927-7.00003-0.
Gottlieb, Scott. “Breaking Down Barriers between Clinical T rials and
Clinical Care: Incorporating Real World Evidence into Regulatory
Decision Making.” Speech to the Bipartisan Policy Center. US Food and
Drug Administration, January 28, 2019. https://w ww.fda.gov/news
-events/speeches-fda-officials/ breaking-down-barriers-between-clinical
-trials-and-clinical-care-incorporating-real-world-evidence.
Haimovitch, Larry, and Richard Mark Kirkner. “A Look Behind Revision
Optics’ Shuttered Door.” Ophthalmology Innovation Source (OIS),
February 14, 2018. https://ois.net/ look-behind-revision-optics-shuttered
-door.
Hajat, Cother, and Emma Stein. “The Global Burden of Multiple Chronic
Conditions: A Narrative Review.” Preventive Medicine Reports 12 (2018):
284–93. https://doi.org/10.1016/j.pmedr.2018.10.008.
286 Bibliography

Harris, Sophie, Ari Miller, Stephanie Amiel, and Henrietta Mulnier.
“Characterization of Adults with Type 1 Diabetes Not Attending
Self-Management Education Courses: The Barriers to Uptake of Type 1
Diabetes Education (BUD1E) Study.” Qualitative Health Research 29, no. 8
(2019): 1174–85. https://doi.org/10.1177/1049732318823718.
Harris, S. M., P. Shah, H. Mulnier, A. Healey, S. M. Thomas, S. A. Amiel,
and D. Hopkins. “Factors Influencing Attendance at Structured Educa-
tion for Type 1 Diabetes in South London.” Diabetic Medicine 34, no. 6
(2017): 828–33. https://doi.org/10.1111/dme.13333.
Hartvigsen, Jan, Mark J. Hancock, Alice Kongsted, Quinette Louw,
Manuela L. Ferreira, Stéphane Genevay, Damian Hoy, et al. “What Low
Back Pain Is and Why We Need to Pay Attention.” Lancet 391, no. 10137
(2018): 2356–67. https://doi.org /10.1016/S0140- 6736(18)30480-X .
Health Sciences Authority. Regulatory Guidelines for Software Medical Devices:
A Lifecycle Approach. Singapore: Health Sciences Authority, December 2019.
https://w ww.hsa.gov.sg/docs/default-source/announcements/regulatory
-updates/regulatory-guidelines-for-software-medical-devices--a-lifecycle
-approach.pdf.
Hehner, Steffen, Stefan Biesdorf, and Manuel Möller. “Digitizing Healthcare—
Opportunities for Germany.” McKinsey & Company, October 31, 2018.
https://w ww.mckinsey.com/industries/ healthcare-systems-and-services
/our-insights/digitizing-healthcare-opportunities-for-germany.
Henze, Stephanie, Amy Hung, Tobias Silberzahn, and Dandi Zhu. “Moving
Digital Health Forward: Lessons on Business Building.” McKinsey &
Company, January 13, 2021. https://w ww.mckinsey.com/industries/ life
-sciences/our-insights/moving-digital-health-forward-lessons-on
-business-building.
Heuvel, Roger van den, Nicolas G. Stamminger, Peter Mozerov, and Girisha
Fernando. Pathway to Success in Outcome-Based Contracting. Zur ich:
KPMG Switzerland, 2020. https://assets.kpmg/content/dam/ kpmg/ch
/pdf/pathway-to-success-in-outcome-based-contracting.pdf.
Hood, Korey K., Marisa Hilliard, Gretchen Piatt, and Carolyn E. Ievers-
Landis. “Effective Strategies for Encouraging Behavior Change in People
with Diabetes.” Diabetes Management (London) 5, no. 6 (2015): 499–510.
https://w ww.ncbi.nlm.nih.gov/pmc/articles/PMC6086609.
Hoult, Molly. “Key F actors to Obtaining Reimbursement for Digital
Therapeutics in the US.” Eversana, February 4, 2021. https://w ww
.eversana.com/insights/ key-factors-to-obtaining-reimbursement-for
-digital-therapeutics-in-the-u-s.
Bibliography 287

Howarth, Ana, Jose Quesada, Jessica Silva, Stephanie Judycki, and Peter R.
Mills. “The Impact of Digital Health Interventions on Health-Related
Outcomes in the Workplace: A Systematic Review.” Digital Health 4
(2018). https://doi.org /10.1177/2055207618770861.
Huber, Stephan, Janosch A. Priebe, Kaja-Maria Baumann, Anne Plidschun,
Christine Schiessl, and Thomas R. Tölle. “Treatment of Low Back Pain
with a Digital Multidisciplinary Pain Treatment App: Short-Term
Results.” JMIR Rehabilitation and Assistive Technologies 4, no. 2 (2017):
e11. https://doi.org /10.2196/rehab.9032.
Hult, Kristopher J. Measuring the Potential Health Impact of Personalized
Medicine: Evidence from Multiple Sclerosis Treatments. Chicago: University
of Chicago Press, 2019.
Humphreys, Gary. “Checklists Save Lives.” Bulletin of the World Health
Organization 86, no. 7 (2008): 501–2. https://doi.org/10.2471/BLT.08
.010708.
Hutchinson, Karen, Regina Sloutsky, Ashley Collimore, Benjamin Adams,
Brian Harris, Terry D. Ellis, and Louis N. Awad. “A Music-Based Digital
Therapeutic: Proof-of-Concept Automation of a Progressive and Individ-
ualized Rhythm-Based Walking Training Program after Stroke.” Neu-
rorehabilitation and Neural Repair 34, no. 11 (2020): 986–96. https://doi
.org /10.1177/1545968320961114.
Iacoviello, Brian M., Joshua R. Steinerman, David B. Klein, Theodore L.
Silver, Adam G. Berger, Sean X. Luo, and Nicholas J. Schork. “Clickotine,
a Personalized Smartphone App for Smoking Cessation: Initial Evalua-
tion.” JMIR mHealth and uHealth 5, no. 4 (2017): e56. https://doi.org/10
.2196/mhealth.7226.
Institute for Clinical and Economic Review. “Cost-Effectiveness, the
QALY, and the evLYG.” Accessed November 1, 2021. https://icer.org
/our-approach/methods-process/cost-effectiveness-t he-qaly-a nd-t he
- evlyg.
Institute of M ental Health. “Singapore Residents Show a High Recognition
of Diabetes, with Older P eople Being More Knowledgeable about the
Condition.” Media release, April 20, 2021. https://w ww.imh.com.sg
/Newsroom/News-Releases/Documents/20Apr2021_ Singapore resi-
dents show a high recognition of diabetes.pdf.
Jacob, Christine, Antonio Sanchez-Vazquez, and Chris Ivory. “Social,
Organizational, and Technological Factors Impacting Clinicians’
Adoption of Mobile Health Tools: Systematic Literature Review.” JMIR
mHealth and uHealth 8, no. 2 (2020). https://doi.org/10.2196/15935.
288 Bibliography

Japsen, Bruce. “Inhaled Insulin’s Cost May Take Breath Away.” Chicago
Tribune, February 1, 2006. https://w ww.chicagotribune.com/news/ct
-x pm-2006-02-02-0602020156-story.html.
Johnson, Andrea. “Homepage Optimization.” Marketing Experiments,
March 23, 2016. https://marketingexperiments.com/a-b -testing/ how
-humana-optimized-banners.
Joyce, Rhod. NHS Digital Health Technology Standard. United Kingdom:
National Health Service (NHS), February 24, 2020. https://assets.nhs.uk
/prod/documents/NHS_Digital_Health_Technology_ Standard_draft.pdf.
Kaplan, Warren, Veronika J. Wirtz, Aukje Mantel-Teeuwisse, Pieter Stolk,
Béatrice Duthey, and Richard Laing. Priority Medicines for Europe and the
World: 2013 Update. Geneva: World Health Organization, 2013. https://
www.researchgate.net/publication/249995018_ Priority_ Medicines_for
_Europe_ and_the_World_2013_Update_ Report.
Kasztura, Miriam, Aude Richard, Nefti-Eboni Bempong, Dejan Loncar, and
Antoine Flahault. “Cost-Effectiveness of Precision Medicine: A Scoping
Review.” International Journal of Public Health 64 (December 2019):
1261–71. https://doi.org /10.1007/s00038-019-01298-x.
Kee, Theodore, Chee Weiyan, Agata Blasiak, Peter Wang, Jordan K. Chong,
Jonna Chen, B. T. Thomas Yeo, Dean Ho, and Christopher L. Asplund.
“Harnessing CURATE.AI as a Digital Therapeutics Platform by Identify-
ing N-of-1 Learning Trajectory Profiles.” Advanced Therapeutics 2, no. 9
(2019): 1900023. https://doi.org /10.1002/adtp.201900023.
Kennedy-Martin, Tessa, Kristina S. Boye, and Xiaomei Peng. “Cost of
Medication Adherence and Persistence in Type 2 Diabetes Mellitus: A
Literature Review.” Patient Preference and Adherence 11 (2017): 1103–17.
https://doi.org/10.2147/PPA.S136639.
Kennedy-Martin, Tessa, Sarah Curtis, Douglas Faries, Susan Robinson, and
Joseph Johnston. “A Literature Review on the Representativeness of
Randomized Controlled Trial Samples and Implications for the External
Validity of Trial Results.” Trials 16, no. 1 (2015): 1–14. https://doi.org/10
.1186/s13063-015-1023-4.
Keown, Alex. “Sienna Biopharmaceuticals Files for Bankruptcy, Delaying
Phase III Psoriasis Trial.” BioSpace, September 18, 2019. https://w ww
.biospace.com/article/sienna-biopharmaceuticals-fi les-chapter-11
-delaying-phase-iii-psoriasis-trial.
Kerr, Ronny. “When Omada Health Was Young: The Early Years.” Vator TV,
March 21, 2017. https://vator.t v/news/2017-03-21-when-omada-health
-was-young-the-early-years.
Bibliography 289

Kesavadev, Jothydev, Ashok Kumar Das, Ranjit Unnikrishnanlst, Sha-
shank R. Joshi, Ambady Ramachandran, Jisha Shamsudeen, Gopika
Krishnan, Sunitha Jothydev, and Viswanathan Mohan. “Use of Insulin
Pumps in India: Suggested Guidelines Based on Experience and Cultural
Differences.” Diabetes Technology & Therapeutics 12, no. 10 (2010):
823–31. https://doi.org/10.1089/dia.2010.0027.
Kim, W. Chan, and Renée Mauborgne. Blue Ocean Strategy: How to Create
Uncontested Market Space and Make the Competition Irrelevant. Boston:
Harvard Business Review Press, 2015.
King, Kathryn M., Philip J. King, Rahul Nayar, and Scott Wilkes. “Percep-
tions of Adolescent Patients of the ‘Lived Experience’ of Type 1 Diabe-
tes.” Diabetes Spectrum 30, no. 1 (2017): 23–35. https://doi.org/10.2337
/ds15-0041.
Klonoff, David C., and David Kerr, “Overcoming Barriers to Adoption of
Digital Health Tools for Diabetes.” Journal of Diabetes Science and
Technology 12, no. 1 (2018): 3–6. https://doi.org/10.1177/193229
6817732459.
Koh, Dean. “KT Signs MoU with Sibel Health to Expand Its Digital Health
Capabilities.” Mobihealth News, November 12, 2020. https://w ww
.mobihealthnews.com/news/asia/ kt-signs-mou-sibel-health-expand-its
-digital-health-capabilities.
Krist, Alex H., Sebastian T. Tong, Rebecca A. Aycock, and Daniel R. Longo.
“Engaging Patients in Decision-Making and Behavior Change to
Promote Prevention.” Studies in Health Technology and Informatics 240
(2017): 284–302.
KT Corporation. “S.Korea’s KT Corp. to Expand Digital Health Business
with Sibel.” November 10, 2020. https://corp.kt.com/eng/ html/promote
/news/report_detail.html?rows=10&page=1&datNo=16036.
Kyle Blankenship. “Camera Maker Kodak Dives into Drug Manufacturing
with $765M Federal Loan.” Fierce Pharma, July 29, 2020. https://w ww
.fiercepharma.com/manufacturing/camera-maker-kodak-dives-into
-drug-manufacturing-765m-federal-loan.
Landi, Heather. “Teladoc’s Virtual Visits Reach 3M during Q4 as Revenue
Grows to $383M.” Fierce Pharma, February 25, 2021. https://w ww
.fiercehealthcare.com/tech/teladoc-s-v irtual-v isits-reach-3m-during-q4
-as-revenue-grows-to-383m.
Landi, Heather. “Telehealth Leader Teladoc to Buy Livongo in $18.5B Deal.”
Fierce Healthcare, August 5, 2020. https://w ww.fiercehealthcare.com
/tech/teladoc-livongo-plan-to-merge-18-5b-deal.
290 Bibliography

Lerner, Jennifer S., Ye Li, Piercarlo Valdesolo, and Karim S. Kassam.
“Emotion and Decision Making.” Annual Review of Psychology 66 (Janu-
ary 2015): 799–823. https://doi.org/10.1146/annurev-psych-010213
-115043.
Levins, Hoag. “Report from the First National ‘Nudge Units in Health Care’
Symposium.” University of Pennsylvania, Leonard Davis Institute of
Health Economics, October 3, 2018. https://ldi.upenn.edu/our-work
/research-updates/report-from-the-first-national-nudge-units-in-health
-care-symposium.
Liu, Xiaoxuan, Livia Faes, Aditya U. Kale, Siegfried K. Wagner, Dun Jack
Fu, Alice Bruynseels, Thushika Mahendiran, et al. “A Comparison of
Deep Learning Performance against Health-Care Professionals in
Detecting Diseases from Medical Imaging: A Systematic Review and
Meta-A nalysis.” Lancet Digital Health 1, no. 6 (2019): e271–97. https://doi
.org/10.1016/S2589-7500(19)30123-2.
Livongo Health. “Livongo Reports Third Quarter 2020 Financial Results.”
Globe Newswire, October 28, 2020. https://w ww.globenewswire.com
/news-release/2020/10/28/2116285/0/en/Livongo-Reports-T hird
-Quarter-2020-Financial-Results.html.
Livongo Health. “United States Securities and Exchange Commission Form
S-1 Registration Statement.” Filed on June 28, 2019. Registration
No. 333-. https://w ww.sec.gov/A rchives/edgar/data/1639225/00011
9312519185159/d731249ds1.htm.
Lorenzoni, Luca, Alberto Marino, David Morgan, and Chris James. Health
Spending Projections to 2030: New Results Based on a Revised OECD
Methodology. OECD Health Working Papers No. 110. Paris: OECD
Publishing, 2019. https://doi.org/10.1787/5667f23d-en.
Luo, Candice, Nitika Sanger, Nikhita Singhal, Kaitlin Pattrick, Ieta Shams,
Hamnah Shahid, Peter Hoang, et al. “A Comparison of Electronically
Delivered and Face-to-Face Cognitive Behavioural Therapies in Depres-
sive Disorders: A Systematic Review and Meta-A nalysis.” eClinical
Medicine 24 (2020). https://doi.org/10.1016/j.eclinm.2020.100442.
Manta, Christine, Bray Patrick-Lake, and Jennifer C. Goldsack. “Digital
Measures that Matter to Patients: A Framework to Guide the Selection
and Development of Digital Measures of Health.” Digital Biomarkers 4,
no. 3 (2020): 69–77. https://doi.org/10.1159/000509725.
Marshall, Sheila K. “Do I M atter? Construct Validation of Adolescents’
Perceived Mattering to Parents and Friends.” Journal of Adolescence 24,
no. 4 (2001): 473–90. https://doi.org/10.1006/jado.2001.0384.
Bibliography 291

Marshall, Sheila K., and Lauree Tilton-Weaver. “Adolescents’ Perceived
Mattering to Parents and Friends: Testing Cross-Lagged Associations
with Psychosocial Well-Being.” International Journal of Behavioral
Development 43, no. 6 (2019): 541–52. https://doi.org/10.1177
/0165025419844019.
Martino, Maureen, Liz Jones Hollis, and Erica Teichert. “Pharma’s Biggest
Flops.” Fierce Pharma, October 18, 2020. https://w ww.fiercepharma.com
/special-report/exubera-pharma-s-biggest-flops.
Matera, Camilla, Nicolina Bosco, and Patrizia Meringolo. “Perceived
Mattering to F amily and Friends, Self-Esteem, and Well-Being.” Psychol
ogy, Health & Medicine 25, no. 5 (2020): 550–58. https://doi.org/10.1080
/13548506.2019.1626454.
Mathews, Anna W., and Danny Yadron. “Health Insurer Anthem Hit by
Hackers.” Wall Street Journal, February 4, 2015. https://w ww.wsj.com
/articles/ health-insurer-anthem-hit-by-hackers-1423103720.
McCarthy, Owen. “Looking Back & Ahead at the Digital Therapeutics
Industry.” Medium, December 23, 2020. https://medrhythms.medium
.com/ looking-back-ahead-at-the-digital-therapeutics-industry
-2e8454d342c5.
McGlynn, Elizabeth A., Steven M. Asch, John Adams, Joan Keesey,
Jennifer Hicks, Alison DeCristofaro, and Eve A. Kerr. “The Quality of
Health Care Delivered to Adults in the United States.” New England
Journal of Medicine 348, no. 26 (2003): 2635–45. https://doi.org/10.1056
/NEJMsa022615.
McGrath, Rita Gunter, and Ian C. MacMillan. “Discovery-Driven Plan-
ning.” Harvard Business Review, July–August 1995. https://hbr.org/1995
/07/discovery-driven-planning.
McIntyre, Harold David, “Dose Adjustment for Normal Eating: A Role for
the Expert Patient?” Diabetes & Metabolism Journal 38, no. 2 (2014):
87–91. https://doi.org /10.4093/dmj.2014.38.2.87.
MedRhythms. “MedRhythms Announces an Expansion of Intellectual
Property Portfolio, Securing Patent on Audio Engine for Digital Thera-
peutics Platform.” Cision PR Newswire, April 17, 2020. https://w ww
.prnewswire.com/news-releases/medrhythms-announces-an-expansion
-of-intellectual-property-portfolio-securing-patent-on-audio-engine
-for-digital-therapeutics-platform-301042485.html.
“Medtronic Profit Margin 2006–2021.” Macrotrends, accessed July 31, 2021.
https://w ww.macrotrends.net/stocks/charts/MDT/medtronic/profit
-margins.
292 Bibliography

Megginson, Leon C. “Lessons from Europe for American Business.” South-
western Social Science Quarterly 44, no. 1 (1963): 3–13. http://w ww.jstor
.org/stable/42866937.
Mehra, Mandeep R., Sapan S. Desai, Frank Ruschitzka, and Amit N. Patel.
“Retracted: Hydroxychloroquine or Chloroquine with or without a
Macrolide for Treatment of COVID-19: A Multinational Registry Analysis.”
Lancet, May 22, 2020. https://doi.org/10.1016/S0140-6736(20)31180-6.
Michie, Susan, Robert West, Kate Sheals, and Cristina A. Godinho. “Evaluat-
ing the Effectiveness of Behavior Change Techniques in Health-Related
Behavior: A Scoping Review of Methods Used.” Translational Behavioral
Medicine 8, no. 2 (2018): 212–24. https://doi.org/10.1093/tbm/ibx019.
Miliard, Mike. “AMA Sees Surge in Health IT Adoption, ‘Rise of the
Digital-Native Physician.’ ” Healthcare IT News, February 6, 2020.
https://w ww.healthcareitnews.com/news/ama-sees-surge-health-it
-adoption-r ise-digital-native-physician.
Mobiquity Inc. “COVID-19: Ensuring a Quality Patient Experience with the
Rise of Digitisation in a Healthcare Setting.” Mobiquity, January 21,
2021. https://w ww.mobiquity.com/insights/covid-19-research-study
-digital-health-report-2021.
Moller, Arlen C., Gina Merchant, David E. Conroy, Robert West, Eric
Hekler, Kari C. Kugler, and Susan Michie. “Applying and Advancing
Behavior Change Theories and Techniques in the Context of a Digital
Health Revolution: Proposals for More Effectively Realizing Untapped
Potential.” Journal of Behavioral Medicine 40 (February 2017): 85–98.
https://doi.org/10.1007/s10865-016-9818-7.
Muoio, Dave. “BlueStar’s Eighth FDA Clearance Adds Long-Acting Basal
Insulin Support for Type 2 Patients.” Mobihealth News, June 3, 2020.
https://w ww.mobihealthnews.com/news/ bluestars-eighth-fda
-clearance-adds-long-acting-basal-insulin-support-t ype-2-patients.
Muoio, Dave. “Proteus Parts Ways with Otsuka as It Pivots toward Oncol-
ogy, Infectious Disease Treatment Adherence.” Mobihealth News,
January 14, 2020. https://w ww.mobihealthnews.com/news/proteus
-parts-ways-otsuka-it-pivots-toward-oncology-infectious-disease
-treatment-adherence.
Murray, Elizabeth, Eric B. Hekler, Gerhard Andersson, Linda M. Collins,
Aiden Doherty, Chris Hollis, Daniel E. Rivera, Robert West, and Jer-
emy C. Wyatt. “Evaluating Digital Health Interventions: Key Questions
and Approaches.” American Journal of Preventive Medicine 51, no. 5
(2016): 843–51. https://doi.org/10.1016/j.amepre.2016.06.008.
Bibliography 293

Naser, Abdallah Y., Anas Nawfal Hameed, Nour Mustafa, Hassan Alwafi,
Eman Zmaily Dahmash, Hamad S. Alyami, and Haya Khalil. “Depres-
sion and Anxiety in Patients With Cancer: A Cross-Sectional Study.”
Frontiers in Psychology 12 (2021): 585534. https://doi.org/10.3389/f psyg
.2021.585534.
National Cancer Institute at the National Institutes of Health. “Standard of
Care.” 2021. https://w ww.cancer.gov/publications/dictionaries/cancer
-terms/def/standard-of-care.
National Center for Advancing Translational Sciences. “About NCATS.”
Accessed September 23, 2021. https://ncats.nih.gov/about.
National Human Genome Research Institute. “Rare Diseases FAQ.” Accessed
January 26, 2021. https://w ww.genome.gov/FAQ/R are-Diseases.
National Institute for Health and Care Excellence (NICE). Continuous
Subcutaneous Insulin Infusion for the Treatment of Diabetes Mellitus:
Technology Appraisal Guidance. London: NICE, 2008. https://w ww.nice
.org.uk/g uidance/ TA151.
National Institute for Health and Care Excellence (NICE). Evidence Standards
Framework for Digital Health Technologies. London: NICE, March 2019.
https://w ww.nice.org.uk/Media/Default/About/what-we-do/our
-programmes/evidence-standards-framework/digital-evidence
-standards-framework.pdf.
National Medical Research Council. Diabetes Taskforce Report. Singapore:
Ministry of Health, 2020. https://w ww.nmrc.gov.sg/docs/default-source
/about-us-library/dtf-summary-report.pdf.
Naujokaite, Ruta. “What is the Digital Care Act (DVG)?” Chino.io Blog, June 23,
2020. https://w ww.chino.io/ blog/what-is-the-digital-care-act-dvg.
Noell, Guillaume, Rosa Faner, and Alvar Agustí. “From Systems Biology to
P4 Medicine: Applications in Respiratory Medicine.” European Respira-
tory Review 27, no. 147 (2018). https://doi.org/10.1183/16000617.0110
-2017.
OCEBM Levels of Evidence Working Group. “The Oxford 2011 Levels of
Evidence.” Centre for Evidence-Based Medicine (CEBM). Accessed
November 17, 2022. https://w ww.cebm.ox.ac.uk/resources/ levels-of
-evidence/ocebm-levels-of-evidence.
Omada Health. “Omada Digital Diabetes Prevention Program Shows
Sustained HbA1c Reduction and Weight Loss in Randomized Control
Trial.” Cision PR Newswire, November 3, 2020. https://w ww.prnewswire
.com/news-releases/omada-digital-diabetes-prevention-program-shows
294 Bibliography

-sustained-hba1c-reduction-and-weight-loss-in-randomized-control
-trial-301165318.html.
Organisation for Economic Co-operation and Development (OECD).
“Employment Rate.” OECD Data, accessed November 17, 2022. https://
data.oecd.org /emp/employment-rate.htm.
Organisation for Economic Co-operation and Development (OECD). Health
at a Glance 2011: OECD Indicators. Paris: OECD Publishing, 2011. https://
www.oecd.org /els/ health-systems/49105858.pdf.
Organisation for Economic Co-operation and Development (OECD).
Out-of-Pocket Spending: Access to Care and Financial Protection. Paris:
OECD, April 2019. https://w ww.oecd.org/ health/ health-systems/OECD
-Focus-on- Out-of-Pocket-Spending-April-2019.pdf.
Pantuck, Allan J., Dong-Keun Lee, Theodore Kee, Peter Wang, Sanjay
Lakhotia, Michael H. Silverman, Colleen Mathis, et al. “Modulating BET
Bromodomain Inhibitor ZEN-3694 and Enzalutamide Combination
Dosing in a Metastatic Prostate Cancer Patient Using CURATE.AI, an
Artificial Intelligence Platform.” Advanced Therapeutics 1, no. 6 (2018).
https://doi.org/10.1002/adtp.201800104.
Parcher, Benjamin, and Megan Coder. “Decision Makers Need an Approach
to Determine Digital Therapeutic Product Quality, Access, and Appro-
priate Use.” Journal of Managed Care & Specialty Pharmacy 27, no. 4
(2021): 536–38. https://doi.org /10.18553/jmcp.2021.27.4.536.
Pear Therapeutics. “Clinical Evidence.” reSET-O and reSET Proven Out-
comes, 2021. https://resetforrecovery.com/outcomes/.
Pear Therapeutics. “Pear Therapeutics Closes $20 Million Financing.” Press
release, accessed November 17, 2022. https://peartherapeutics.com/pear
-therapeutics-raises-20-million-to-launch-a-family-of-prescription
-digital-therapies-treating-disorders-of-the-brain.
Pear Therapeutics. “Reset for Recovery: The Proof.” Accessed November, 11,
2022. https://w ww.resetforrecovery.com/reset#the-proof.
Peyrot, Mark, R. R. Rubin, T. Lauritzen, F. J. Snoek, D. R. Matthews, and
S. E. Skovlund. “Psychosocial Problems and Barriers to Improved
Diabetes Management: Results of the Cross-National Diabetes Atti-
tudes, Wishes and Needs (DAWN) Study.” Diabetic Medicine 22, no. 10
(2005): 1379–85. https://doi.org/10.1111/j.1464-5491.2005.01644.x.
Pichai, Sundar. “Privacy Should Not Be a Luxury Good.” New York
Times, May 7, 2019. https://w ww.nytimes.com/2019/05/07/opinion
/google-sundar-pichai-privacy.html.
Bibliography 295

Piller, Charles, and Kelly Servick. “Two Elite Medical Journals Retract
Coronavirus Papers over Data Integrity Questions.” Science, June 4,
2020. https://w ww.science.org/news/2020/06/t wo-elite-medical
-journals-retract-coronavirus-papers-over-data-integrity-questions.
Ping An. 2020 Annual Report. Shenzhen: Ping An, February 3, 2021.
https://group.pingan.com/resource/pingan/IR-Docs/2020/pingan-ar20
-report.pdf.
Ping An. Announcement of Unaudited Results for the Three Months Ended
March 31, 2020. Shenzhen: Ping An, April 23, 2020. https://doc.irasia
.com/ listco/ hk/pingan/announcement/a228455-e_02318ann
_20200423(20200423_1445).pdf.
Ping An. “Ping An Good Doctor Has Become the First Online Healthcare
Platform with More Than 300 Million Registered Users.” Cision PR
Newswire, September 23, 2019. https://w ww.prnewswire.com/news
-releases/ping-an-good-doctor-has-become-the-first-online-healthcare
-platform-with-more-than-300-million-registered-users-300923026.html.
Ping An. “Ping An Unveils Health Care Ecosystem Strategy.” Cision PR News-
wire, September 23, 2020. https://w ww.prnewswire.com/news-releases
/ping-an-unveils-health-care-ecosystem-strategy-301136665.html.
Piper, Megan E., Michael C. Fiore, Stevens S. Smith, David Fraser, Dan-
iel M. Bolt, Linda M. Collins, Robin Mermelstein, et al. “Identifying
Effective Intervention Components for Smoking Cessation: A Factorial
Screening Experiment.” Addiction 111, no. 1 (2016): 129–41. https://doi
.org/10.1111/add.13162.
Plsek, Paul. “Redesigning Health Care with Insights from the Science of
Complex Adaptive Systems.” In Crossing the Quality Chasm: A New Health
Care System for the 21st C entury, edited by Institute of Medicine Commit-
tee on Quality of Health Care in America, 309–22. Washington, DC:
National Academy Press, 2001.
Prabhu, Nishitha, and Parvathy R. Menon. “Express Scripts a Pioneer:
Launches Curated List of Digital Therapeutics.” Clarivate, December 12,
2019. https://clarivate.com/ blog/express-scripts-pioneer-launches
-curated-list-digital-therapeutics.
Precedence Research. “Digital Therapeutics Market Size to Hit US$ 11.82
Bn by 2027.” Globe Newswire, October 19, 2020. https://w ww.globe
newswire.com/news-release/2020/10/19/2110561/0/en/Digital
-T herapeutics-Market-Size-to-Hit-US-11-82-Bn-by-2027.html.
Priebe, Janosch A., Katharina K. Haas, Leida F. Moreno Sanchez, Karin
Schoefmann, Daniel A. Utpadel-Fischler, Paul Stockert, Reinhard
296 Bibliography

Thoma, et al. “Digital Treatment of Back Pain versus Standard of Care:
The Cluster-R andomized Controlled Trial, Rise-uP.” Journal of Pain
Research 13 (2020): 1823–38. https://doi.org/10.2147/JPR.S260761.
Ravot, Elisabetta, and Roberto Ascione. “Access/Reimbursement Policies
for Digital Therapeutics Already in Use in National Health Systems.”
Tendenze nuove, Special Issue 4 (2021): 105–16. https://w ww.tenden
zenuove.it/2021/07/29/access-reimbursement-policies-for-digital
-therapeutics-a lready-in-use-in-national-health-systems.
Resnikoff, Serge, Van Charles Lansingh, Lindsey Washburn, William Felch,
Tina-Marie Gauthier, Hugh R. Taylor, Kristen Eckert, David Parke, and
Peter Wiedemann. “Estimated Number of Ophthalmologists Worldwide
(International Council of Ophthalmology Update): W ill We Meet the
Needs?” British Journal of Ophthalmology 104, no. 4 (2020): 588–92.
https://doi.org /10.1136/ bjophthalmol-2019-314336.
Reuter, Elise. “Proteus Files for Bankruptcy: Where Did It Falter?” MedCity
News, June 16, 2020. https://medcitynews.com/2020/06/proteus-fi les
-for-bankruptcy-where-did-it-falter.
Riddle, Matthew C., Julio Rosenstock, and John Gerich, “The Treat-to-
Target Trial: Randomized Addition of Glargine or Human NPH Insulin
to Oral Therapy of Type 2 Diabetic Patients.” Diabetes Care 26, no. 11
(2003): 3080–86. https://doi.org /10.2337/diacare.26.11.3080.
Robbins, Michele, Janice Tufte, and Clar issa Hsu. “Learning to ‘Swim’
with the Experts: Experiences of Two Patient Co-Investigators for a
Project Funded by the Patient-Centered Outcomes Research Insti-
tute.” Permanente Journal 20, no. 2 (2016): 85–88. https://doi.org /10
.7812/TPP/15-162.
Roehl, Michael Joseph. “DiGA: Digital Therapeutics in Germany.” LinkedIn,
March 9, 2022. https://w ww.linkedin.com/pulse/diga-digital-therapeutics
-germany-michael-joseph-roehl.
Ross, Jamie, Fiona Stevenson, Rosa Lau, and Elizabeth Murray. “Factors
that Influence the Implementation of e-Health: A Systematic Review of
Systematic Reviews (An Update).” Implementation Science 11, no. 1 (2016):
1–12. https://doi.org/10.1186/s13012-016-0510-7.
Rossignol, S., and G. Melvill Jones. “Audio-Spinal Influence in Man Studied
by the H-Reflex and Its Possible Role on Rhythmic Movements Synchro-
nized to Sound.” Electroencephalography and Clinical Neurophysiology 41,
no. 1 (1976): 83–92. https://doi.org/10.1016/0013-4694(76)90217-0.
Rubin, Devon I. “Epidemiology and Risk Factors for Spine Pain.” Neurologic
Clinics 25, no. 2 (2007): 353–71. https://doi.org/10.1016/j.ncl.2007.01.004.
Bibliography 297

Sanofi. “Sanofi and Abbott Partner to Integrate Glucose Sensing and
Insulin Delivery Technologies to Help Change the Way Diabetes Is
Managed.” Press release, September 16, 2019. https://w ww.sanofi.com
/en/media-room/press-releases/2019/2019-09-16-07-00-00.
Savonix, “Savonix and Boston University School of Public Health Launch
Landmark Population Health Alzheimer’s Disease Discovery Study
(ASSIST) to Digitally Collect Brain Health Data.” Businesswire, Novem-
ber 04, 2019. https://w ww.businesswire.com/news/ home
/20191104005287/en/Savonix-and-Boston-University-School-of-Public
-Health-Launch-Landmark-Population-Health-A lzheimer’s-Disease
-Discovery-Study-A SSIST-to-Digitally- Collect-Brain-Health-Data.
Savnoix. “Savonix and Fujitsu Connected Technologies Limited Partner to
Introduce Savonix Mobile App on Raku-R aku Smartphone F-42A.”
Businesswire, September 29, 2020. https://w ww.businesswire.com/news
/home/20200929005016/en/Savonix-and-Fujitsu- Connected-Techno
logies-Limited-Partner-to-Introduce-Savonix-Mobile-App-on-R aku
-R aku-Smartphone-F-42A.
“Scailyte among the Spinoff Finalists for the Global Nature Research
Awards.” Swiss Biotech, June 20, 2020. https://w ww.swissbiotech.org
/listing/scailyte-among-the-spinoff-prize-finalists.
Shah, Nigam H., Arnold Milstein, and Steven C. Bagley. “Making Machine
Learning Models Clinically Useful.” JAMA 322, no. 14 (2019): 1351–52.
https://doi.org/10.1001/jama.2019.10306.
Sham, Julian. Accenture 2019 Digital Health Consumer Survey: Singapore
Results. Accenture Health, 2019. https://w ww.accenture.com/_ acnmedia
/pdf-102/accenture-digital-health-consumer-survey-singapore.pdf.
Shen, Yinzhong, Tingyi Liu, Jun Chen, Xin Li, Li Liu, Jiayin Shen, Jian-
grong Wang, et al. “Harnessing Artificial Intelligence to Optimize
Long-Term Maintenance Dosing for Antiretroviral-Naive Adults with
HIV-1 Infection.” Advanced Therapeutics 3, no. 4 (2019): 1900114.
https://doi.org /10.1002/adtp.201900114.
Sherwin, Robert S., Robert M. Anderson, John B. Buse, Marshall H. Chin,
David Eddy, Judith Fradkin, Theodore G. Ganiats, et al. “The Prevention
or Delay of Type 2 Diabetes.” Supplement, Diabetes Care 26, no. S1
(2003): S62–69. https://doi.org /10.2337/diacare.26.2007.S62.
Singapore Ministry of Health. “National Health Survey 2010.” Novem-
ber 2011. https://w ww.moh.gov.sg/resources-statistics/reports/national
-health-survey-2010.
298 Bibliography

Singapore News Center. “Zuellig Pharma Shapes the Future of Healthcare
with the Cloud.” Microsoft, August 26, 2019. https://news.microsoft
.com/en-sg/2019/08/26/zuellig-pharma-shapes-the-f uture-of
-healthcare-w ith-the-cloud/.
Singapore Statutes Online. “Health Products Act 2007.” Accessed March 1,
2021. https://sso.agc.gov.sg/Act/HPA2007.
Sinsky, Christine, Lacey Colligan, Ling Li, Mirela Prgomet, Sam Reynolds,
Lindsey Goeders, Johanna Westbrook, Michael Tutty, and George Blike.
“Allocation of Physician Time in Ambulatory Practice: A Time and
Motion Study in 4 Specialties.” Annals of Internal Medicine 165, no. 11
(2016): 753–60. https://doi.org /10.7326/M16-0961.
Speich, Benjamin, Belinda von Niederhäusern, Nadine Schur, Lars G. Hem-
kens, Thomas Fürst, Neera Bhatnagar, Reem Alturki, et al. “Systematic
Review on Costs and Resource Use of Randomized Clinical T rials Shows a
Lack of Transparent and Comprehensive Data.” Journal of Clinical Epidemi-
ology 96 (2018): 1–11. https://doi.org/10.1016/j.jclinepi.2017.12.018.
States, Rebecca A., Evangelos Pappas, and Yasser Salem. “Overground
Physical Therapy Gait Training for Chronic Stroke Patients with
Mobility Deficits.” Cochrane Database of Systematic Reviews, no. 3 (2009):
CD006075. https://doi.org/10.1002/14651858.CD006075.pub2.
Staton, Tracy. “The Top 10 Patent Losses of 2015.” Fierce Pharma, Decem-
ber 17, 2014. https://w ww.fiercepharma.com/special-report/abilify-0.
Stern, Theodore A., Gregory L. Fricchione, and Jerrold F. Rosenbaum, Mas
sachusetts General Hospital Handbook of General Hospital Psychiatry.
Elsevier Health Sciences, 2010. https://doi.org/10.1016/B978-1-4377
-1927-7.00003-0.
STL Partners. “Four Strategies for Telcos in Healthcare.” Accessed Novem-
ber 17, 2022. https://stlpartners.com/digital-health-telecoms/four
-strategies-for-telcos-in-healthcare.
Subramaniam, M., E. Abdin, L. Picco, S. Pang, S. Shafie, J.A. Vaingankar,
K. W. Kwok, K. Verma, and S. A. Chong. “Stigma t owards People with
Mental Disorders and Its Components: A Perspective from Multi-Ethnic
Singapore.” Epidemiology and Psychiatric Sciences 26, no. 4 (2017): 371–82.
Szajewska, Hania. “Evidence-Based Medicine and Clinical Research: Both
Are Needed, Neither Is Perfect.” Annals of Nutrition and Metabolism 72,
no. suppl_3 (2018): 13–23. https://doi.org/10.1159/000487375.
Teladoc Health. “Teladoc Health and Livongo Merge to Create New Standard
in Global Healthcare Delivery, Access and Experience.” Press release,
Bibliography 299

August 5, 2020. https://w ww.teladochealth.com/newsroom/press
/release/ Teladoc-Health-and-Livongo-Merge-to- Create-New-Standard
-in- Global-Healthcare-Delivery-Access-and-Experience-08-05-2020.
Teladoc Health. “Teladoc Health Reports Third-Quarter 2020 Results.”
Press release, October 28, 2020. https://ir.teladochealth.com/news-and
-events/investor-news/press-release-details/2020/ Teladoc-Health
-Reports-T hird-Quarter-2020-Results/default.aspx.
Teladoc Health. “Teladoc Health to Announce Third Quarter 2022 Finan-
cial Results.” Press release, October 11, 2022. https://ir.teladochealth
.com/news-and-events/investor-news/press-release-details/2022
/Teladoc-Health-to-A nnounce-T hird-Quarter-2022-Financial-Results
/default.aspx.
Thielbar, Kelly O., Kristen M. Triandafilou, Alexander J. Barry, Ning Yuan,
Arthur Nishimoto, Joelle Johnson, Mary Ellen Stoykov, Daria Tsoup-
ikova, and Derek G. Kamper. “Home-Based Upper Extremity Stroke
Therapy Using a Multiuser Virtual Reality Environment: A Randomized
Trial.” Archives of Physical Medicine and Rehabilitation 101, no. 2, 196–203.
https://doi.org/10.1016/j.apmr.2019.10.182.
TM Forum. “Telco Cloud Orchestration Plus, Using Open API on IoT.”
Accessed November 17, 2022. https://w ww.tmforum.org/collaboration
/catalyst-program/telco-cloud-orchestration-plus-using-open-api-iot.
Toelle, Thomas R., Daniel A. Utpadel-Fischler, Katharina-K ristina Haas,
and Janosch A. Priebe. “App-Based Multidisciplinary Back Pain Treat-
ment versus Combined Physiotherapy Plus Online Education: A Ran-
domized Controlled Trial.” NPJ Digital Medicine 2 (May 2019): 1–9.
https://doi.org/10.1038/s41746-019-0109-x.
Toh, Ee Ming, and Wong Pei Ting. “The Big Read: Apathy, Complacency—
the Worst Enemies in Singapore’s War against Diabetes.” Today, Au-
gust 25, 2017. https://w ww.todayonline.com/singapore/apathy
-complacency-worst-enemies-in-war-against-diabetes.
Trimble, Michael, and Dale Hesdorffer. “Music and the Brain: The Neuro-
science of M usic and Musical Appreciation.” BJPsych International 14,
no. 2 (2017): 28–31. https://doi.org/10.1192/s2056474000001720.
Trzeciak, Stephen, and Anthony Mazzarelli. Compassionomics: The Revolu-
tionary Scientific Evidence that Caring Makes a Difference. Pensacola, FL:
Studer Group, 2019.
UCB Biopharma. “Study to Evaluate the Impact of Using Wearable Devices
in Addition to Standard Clinical Practice on Parkinson’s Subject Symp-
300 Bibliography

toms Management.” ClinicalTrials.gov, February 19, 2019. https://
clinicaltrials.gov/ct2/show/NCT03103919.
UpScripts Somryst. “Train Your Brain for Better Sleep.” Accessed Novem-
ber 17, 2022. https://w ww.sleepdocconsult.com/products/somryst-for
-chronic-insomnia.
US Food and Drug Administration. “Artificial Intelligence and Machine
Learning in Software as a Medical Device.” Accessed January 12, 2021.
https://w ww.fda.gov/medical-devices/software-medical-device-samd
/artificial-intelligence-and-machine-learning-software-medical
-device#regulation.
US Food and Drug Administration. “De Novo Classification Request.”
Accessed November 20, 2019. https://w ww.fda.gov/medical-devices
/premarket-submissions/de-novo-classification-request.
US Food and Drug Administration. “Digital Health Software Precertifica-
tion (Pre-Cert) Pilot Program.” Accessed May 6, 2019. https://w ww.fda
.gov/medical-devices/digital-health-center-excellence/digital-health
-software-precertification-pre-cert-program.
US Food and Drug Administration. “Investigational Device Exemption
(IDE).” Accessed December 13, 2019. https://w ww.fda.gov/medical
-devices/ how-study-and-market-your-device/investigational-device
-exemption-ide.
US Food and Drug Administration. “Premarket Approval (PMA).” May 16,
2019. https://w ww.fda.gov/medical-devices/premarket-submissions
/premarket-approval-pma.
US Food and Drug Administration. “Premarket Notification 510(k).”
Accessed March 13, 2020. https://w ww.fda.gov/medical-devices
/premarket-submissions/premarket-notification-510k.
US Food and Drug Administration. Use of Real-World Evidence to Support
Regulatory Decision-Making for Medical Devices: Guidance for Industry and
Food and Drug Administration Staff. Silver Spring, MD: FDA, August 31,
2017. https://w ww.fda.gov/media /99447/download.
Van Oostrom, Sandra H., H. Susan J. Picavet, Simone R. de Bruin, Irina
Stirbu, Joke C. Korevaar, Francois G. Schellevis, and Caroline A. Baan.
“Multimorbidity of Chronic Diseases and Health Care Utilization in
General Practice.” BMC F amily Practice 15, no. 1 (2014): 1–9. https://doi
.org/10.1186/1471-2296-15- 61.
Verghese, Joe, Roee Holtzer, Richard B. Lipton, and Cuiling Wang. “Quanti-
tative Gait Markers and Incident Fall Risk in Older Adults.” Journals of
Bibliography 301

Gerontology: Series A 64A, no. 8 (2009): 896–901. https://doi.org/10.1093
/gerona/glp033.
Vohra, Sunita, Larissa Shamseer, Margaret Sampson, Cecilia Bukutu,
Christopher H. Schmid, Robyn Tate, Jane Nikles, Deborah R. Zucker,
Richard Kravitz, Gordon Guyatt, Douglas G. Altman, and David Moher.
“CONSORT Extension for Reporting N-of-1 Trials (CENT) 2015 State-
ment.” BMJ 350 (2015). https://doi.org/10.1136/ bmj.h1738.
Volk, JoAnn, Dania Palanker, Madeline O’Brien, Christina L. Goe. “States’
Actions to Expand Telemedicine Access during COVID-19 and Future
Policy Considerations.” The Commonwealth Fund, June 23, 2021.
https://w ww.commonwealthfund.org/publications/issue-briefs/2021
/jun/states-actions-expand-telemedicine-access-covid-19.
Voyer, Benjamin G. “ ‘Nudging’ Behaviours in Healthcare: Insights from
Behavioural Economics.” British Journal of Healthcare Management 21,
no. 3 (2015): 130–35. https://doi.org/10.12968/ bjhc.2015.21.3.130.
“VP Live.” Virgin Pulse. Accessed November 17, 2022. https://community
.v irginpulse.com/v p-live-shortlister.
Wei, Khor Ing. “Nurture or Nature? Research Project Reveals Answers.”
MediCine, no. 27 (2018): 18–19. https://medicine.nus.edu.sg/w p-content
/uploads/2018/08/MediCine_ Issue27.pdf.
Welldoc Inc. “Welldoc Receives FDA Clearance for Long-Acting Insulin
Support for Award-Winning Digital Health Solution BlueStar.” Cision PR
Newswire, June 3, 2020. https://w ww.prnewswire.com/news-releases
/welldoc-receives-fda-clearance-for-long-acting-insulin-support-for
-award-w inning-digital-health-solution-bluestar-301069807.html.
Welldoc Inc. “WellDoc Validates Potential of Its Digital Therapeutic,
BlueStar, to Significantly Reduce Healthcare Costs,” Global Newswire,
January 4, 2008. https://w ww.globenewswire.com/en/news-release
/2018/01/04/1283402/36511/en/ WellDoc-Validates-Potential-of-Its
-Digital-T herapeutic-BlueStar-to-Significantly-Reduce-Healthcare- Costs
.html.
Wenger, Nathalie, Marie Méan, Julien Castioni, Pedro Marques-Vidal,
Gérard Waeber, and Antoine Garnier. “Allocation of Internal Medicine
Resident Time in a Swiss Hospital: A Time and Motion Study of Day and
Evening Shifts.” Annals of Internal Medicine 166, no. 8 (2017): 579–86.
https://doi.org /10.7326/M16-2238.
Whaley, Christopher M., Jennifer B. Bollyky, Wei Lu, Stefanie Painter,
Jennifer Schneider, Zhenxiang Zhao, Xuanyao He, Jennal Johnson,
and Eric S. Meadows. “Reduced Medical Spending Associated with
302 Bibliography

Increased Use of a Remote Diabetes Management Program and Lower
Mean Blood Glucose Values.” Journal of Medical Economics 22, no. 9
(2019): 869–77. https://doi.org /10.1080/13696998.2019.1609483.
Wientraub, Arlene. “Pfizer’s Exubera Flop.” Bloomberg, October 18, 2007.
https://w ww.bloomberg.com/news/articles/2007-10-18/pfizers-exubera
-flopbusinessweek-business-news-stock-market-and-financial-advice.
World Health Organization. Adherence to Long-Term Therapies: Evidence for
Action. Geneva: WHO, 2003. https://apps.who.int/iris/ handle/10665
/42682.
World Health Organization. “Asthma: Key Facts.” Accessed March 3, 2021.
https://w ww.who.int/news-room/fact-sheets/detail/asthma.
World Health Organization. “Cardiovascular Diseases (CVDs).” June 11, 2021.
http://w ww.who. int/mediacentre/factsheets/fs317/en/index.html.
World Health Organization. “Dementia: Key Facts.” September 20, 2022.
https://w ww.who.int/news-room/fact-sheets/detail/dementia.
World Health Organization, “Diabetes: Key Facts.” Accessed April 13, 2021.
https://w ww.who.int/news-room/fact-sheets/detail/diabetes.
World Health Organization. Global Strategic Directions for Strengthening
Nursing and Midwifery 2016–2020. Geneva: World Health Organization,
2016. https://apps.who.int/iris/rest/ bitstreams/1157318/retrieve.
World Health Organization. “The Top 10 Causes of Death.” December 9,
2020. https://w ww.who.int/news-room/fact-sheets/detail/the-top-10
-causes-of-death.
World Health Organization. “WHO Director-General’s Opening Remarks
at the Media Briefing on COVID-19.” May 25, 2020. https://w ww.who.int
/director-general/speeches/detail/who-director-general-s-opening
-remarks-at-the-media-briefing-on-covid-19- - -25-may-2020.
Wu, Aimin, Lyn March, Xuanqi Zheng, Jinfeng Huang, Xiangyang Wang,
Jie Zhao, Fiona M. Blyth, Emma Smith, Rachelle Buchbinder, and
Damian Hoy. “Global Low Back Pain Prevalence and Years Lived with
Disability from 1990 to 2017: Estimates from the Global Burden of
Disease Study 2017.” Annals of Translational Medicine 8, no. 6 (2020): 299.
https://doi.org/10.21037/atm.2020.02.175.
Yonhap. “Telcos Expand Digital Health Care Serv ices amid Pandemic.”
Korea Herald, September 14, 2020. http://w ww.koreaherald.com/v iew
.php?ud=20200914000919.
Young, Scott W. H. “Improving Library User Experience with A/B Testing:
Principles and Process.” Weave: Journal of Library User Experience 1, no. 1
(2014). https://doi.org /10.3998/weave.12535642.0001.101.
Bibliography 303

Zaleha H, S., Nora Ithnin, Nur Haliza Abdul Wahab, and Noorhazirah
Sunar. “Intelligent Locking System Using Deep Learning for Autono-
mous Vehicle in Internet of T hings.” International Journal of Advanced
Computer Science and Applications 12, no. 10 (2021): 565–78. http://d x.doi
.org/10.14569/IJACSA.2021.0121063.
Zarrinpar, A., Lee, D. K., Silva, A., Datta, N., Kee, T., Eriksen, C., Weigle, K.,
Agopian, V., Kaldas, F., Farmer, D., Wang, S. E., Busuttil, R., Ho, C. M., &
Ho, D. “Individualizing Liver Transplant Immunosuppression Using a
Phenotypic Personalized Medicine Platform.” Science Translational
Medicine 8, no. 333 (2016). https://doi.org/10.1126/scitranslmed
.aac5954.
Zelinka, Ivan, Thanh Cong Truong, Diep Quoc Bao, Lumir Kojecky, and
Eslam Amer. “Artificial Intelligence in Astrophysics.” In Intelligent
Astrophysics, edited by Ivan Zelinka, Massimo Brescia, and Dalya Baron,
1–28. Cham, Switzerland: Springer Nature, 2021.
304 Bibliography

Index
Abilify (aripiprazole), 193, 194, 201 Anthem Inc., 164

A/B testing, 102–3, 104, 105, 106–8 Apple Inc., 123, 221, 224, 225
Academy for Neurologic M usic Therapy, application programming interfaces
89n1 (APIs), 218
Acceset, 89, 272–73, 274 artificial general intelligence (AGI),
ADHD Insight app, 253 259–60, 262
adoption of DTx solutions: in clinical artificial intelligence (AI): applications
workflows, 176; cost-effectiveness in health care, 7, 140, 259, 265–66;
aspects of, 178–79; decision-making capabilities of, 218, 259–60; categories
processes and, 176; main goals of, 181; of, 259–63; definition of, 259; in digital
nudging principle, 185–88; omnichan- therapeutics, potential for, 266–67,
nel approach to, 181; patient-doctor 268; for drug dosing, 86; functionality
interactions and, 177–78, 180, 182–84; of, 262–63; hierarchy of, 260; public
patients’ involvement in, 176, 179–82, perception of, 177; regulatory chal-
184–87, 188; peer support and, 183–84; lenges of, 267–68; subsets of, 263–65
perspectives on, 173, 188; physicians’ artificial narrow intelligence (ANI), 259,
engagement in, 174, 175, 188; trust in, 260
182–83; value-based medicine and, 179; artificial super intelligence (ASI), 259–60
workflow changes and, 177, 179 Asia Pacific Medical Technology Associa-
Aetna, 244 tion (APACMed), 202
AEvice Health, 104–5, 110, 198 asthma, 105, 115, 202
Akili Interactive: design ethos of, 251–52; attention deficit hyperactivity disorder
DTx solutions, 27, 134, 191, 205, 252, (ADHD), 67, 134, 191, 205, 252, 253
253; establishment of, 249–50; market- augmenting diagnostics, 265
based approach, 255; pharma partner- augmenting drug discovery and develop-
ship, 200; use of video games, 83 ment, 266
Alexa, 260–61 autism, 82
AliHealth, 222
Alzheimer’s disease, 82, 225, 251 Barker, Pierre, 169
Alzheimer’s Disease Discovery Study Behavioural Insights Team (a.k.a. Nudge
(ASSIST), 225 Unit), 185
Amalgam, 200 Belot, Benjamin, 91
Amazon, 221, 223 BenevolentAI, 266
American Medical Collection Agency, 164 Berkshire Hathaway, 223
Amgen, 245 bias in medicine, 120
analgesics, 1 big data, 152n11, 220
Ang, Adrian, 105 Big Health, 175, 182, 183
305

Big Tech companies: channel partner- 200–202, 204–5, 226, 254–55; with
ships, 221–25; health care ventures of, phar m aceut ic al distributors, 203–4;
223–24; marketplaces, 223; philosophy recommendations for, 228–29; with
of “self-serve,” 222 telecommunications companies,
Biofourmis, 32 217–21, 226; with teleX o rganizations,
Biovitals Analytics Engine, 32–33 214–17
BioXcel Therapeutics, 266 chatbots, 261
blood pressure control study, 238–39 ChatGPT, 262
Blue Note Therapeutics, 84–85 checklists, 39
Blue Ocean Strategy, 12, 69 China Mobile, 218
BlueStar®, 24–28, 241 chloroquine, 133
BlueStar’s Insulin Adjustment Program Christensen, Clayton M., 57
(IAP), 26, 87–88 chronic diseases, 3, 4, 104–5, 209
Booty, Edward, 216 chronic obstructive pulmonary disease
brain rehabilitation therapy, 83 (COPD), 202
Breazeal, Cynthia, 263 Cigna, 213, 244
Brocklebank, Nicholas, 122 Cincinnati C hildren’s Hospital, 235
BT Group, 218 Click Therapeutics, 27
business planning tools, 95 click-t hrough rates (CTRs), 106, 107, 108
business-to-business (B2B) model, 73, clinical endpoints, 131
112n7, 197–98, 232, 241 clinical evidence, 29, 256
business-to-business-to-consumer (B2B2C) clinical needs, 10, 41, 77, 88–89
model, 197, 198, 219, 241, 247 clinical t rials: data collection, 146–47,
business-to-consumer (B2C) model, 73, 151; data storage and sharing, 147;
112n7, 197, 198, 219, 232, 241, 247 design and execution of, 131, 151;
business-to-payer model, 232 enrollment practices, 30–31
business ventures, 96, 97, 98, 99 ClinicalTrials.gov, 129
business viability, 91, 95, 96 clinicaltrialsregister.eu, 129
Buyer Utility Map, 64 Clinical Trials Transformation Initiative
buying decision process, 68–73 (CTTI), 133, 145, 148
clinical validation, 113, 130, 136–38, 150,
Cahill, Ken, 31 151
cancer: screening, 218; statistics, 3 cloud platforms, 218
capitation, 223 cognitive behavioral stress management
cardiovascular diseases, 3, 33 (CBSM), 85
care delivery: variability of, 113–15 cognitive behavioral therapy (CBT), 84,
Centers for Medicare & Medicaid 85, 178
Services, 214 cognitive bias, 103
chain of buyers, 69–70, 72–73 compensatory behaviors, 61
channel partnerships: with Big Tech concept tests, 109
companies, 221–25; with employers, Conference on Computer and Communi-
206–10; with insurance companies, cations Security, 125
210–13; key f actors of, 227; with Conference on Neural Information
medical device companies, 202–3; Processing Systems, 125
models of, 229; as path to DTx Connected Clinical Trials (CCT), 146
distribution and commercialization, CONSORT 2010 Statement, 142
10–11, 193, 194, 197, 198–200, 242–43, contract research o rganizations (CROs),
254–55; with pharmaceutical companies, 130
306 Index

Cortana, 260 61; DTx for, 24–25, 56, 58–59, 156–57,
cost-per-c lick, 106 209–10; emotions and pain points
cost-per-completion, 106 of, 64; evaluation of what is right for,
COVID-19: contact tracing apps, 124; 61–62; financial burden, 50; ideal
digital health adoption during, 191–92, journey of, 64; improved outcomes
250–51; virtual care, 180 for, 185; JTBD framework for, 58, 63;
CURATE.AI, 86, 87, 140–42 lifestyle of, 49; long-term and short-
curse of knowledge, 103 term problems, 60; personalized health
program for, 209; relationship with
DarioHealth, 198 physicians, 60; self-management edu-
data, medical: access to, 117–18, 122; cation, 51–52; therapy nonadherence,
analytical tools for, 120–21; biased 53–54
interpretation of, 120, 121; as business Diabetes Prevention Program (DPP), 84,
asset, 126; collection of, 117, 121, 146–47, 213
255; licensed, 122–23; management of, diabetes self-m anagement education
117; ownership of, 117; privacy of, 117, (DSME), 51, 52–53
123–24, 125, 126, 127; provenance of, diabetic retinopathy (DR) screening,
119; retrospective, 118; safety and 158–60
security of, 125, 126, 164; sources of, diagnostic and monitoring, 21, 22, 36n3
117, 118, 119–21, 126; storage and diffusion of innovation, 155
sharing of, 147; types of, 118 digital cognitive behavioral therapy
data commons, 118 (dCBT), 84, 85
data processing products (DPP), 117, 119, digital endpoints (a.k.a. digital biomark-
121–23, 126 ers), 131, 132–33, 151
Dean Foods, 210 digital health: adoption of, 191–92, 228;
decentralized clinical t rials (DTCs), 143, business models, 197–98; categories
144–45 of, 21, 22; data management demands,
decision-making p rocess: biases in, 120, 174; definition of, 21; vs. digital thera-
185 peutics, 21, 24, 32; “post hoc” approach,
Deep Blue, 262 273; regulations, 33, 37n21; technolo-
deep learning (DL), 263, 264, 265 gies, 150, 157; vs. traditional medicine,
dementia: assumptions about patient 253–54
with, 106; cognitive risk assessment, Digital Health Software Precertification
107; DTx for, 107–8; gauging interest (Pre-Cert) Pilot Program, 34, 150
test, 106–7; painted door test, 107–8; Digital Health Technology Standard
screening, 107, 110; statistics, 3 (DHTS), 33–34
De Novo risk-based classification, 148–49 digital literacy, 123
depression, 206, 270–71 Digital Medicine Society (DiMe), 132–33,
diabetes: diagnostics, 43, 48; as global 151
problem, 48; interpretation of data on, digital neurotherapeutics, 83
120; long-term side effects, 60; man- digital placebos, 134, 135
agement of, 44–45, 49, 50, 51, 55, 73, digital therapeutics (DTx): artificial
184–85, 186; public health messages intelligence and, 266–67, 268; attrition
about, 49; statistics, 3, 24, 209; US rates, 108; behavioral change strategies
government spending on, 56 in, 85; benefits of, 10, 23, 47–48, 251–52;
diabetes patients: behavioral change, best practices, 110–11; business models
54–55; challenges of, 48, 54; compen- for, 9, 232, 254–55; business partner-
satory behaviors of, 61; daily routine, ship due diligence, 227; business risks
Index 307

digital therapeutics (DTx) (cont.) Digital Therapeutics Alliance (DTA), 22,
related to, 111n4; capabilities of, 34–35; 266
checklists for, 40; clinical evidence of dilution of control, 96, 99n7
effectiveness of, 28–29, 136; continu- direct-to-consumer distribution model,
ous evolution of, 165–66, 167; cultural 205
setting and, 135–36; data privacy diseases: statistics, 82; top 10 causes of
and, 124; definition of, 3, 22–24, 266; death, 24, 82, 90n13
demand-d riven vs. supply-d riven, 64; disruptive technologies, 224
development of, 135, 226; vs. digital distribution services, 203–4, 205
health solutions, 21, 24, 32; distribu- DKSH, 203
tion strategies for, 194, 197–98, 227; DNA sequencing, 7
economic value of, 193–94; effective- Dose Adjustment for Normal Eating
ness of, 154; emotional proposition, 67; program, 52
evidence-based framework, 98, 270, drugs: dosing methods, 140, 141; vs.
272; evolution of, 250–51, 254; exam- supplements, 31–32
ples of, 27–28; feedback of users, 252; Drugstore.com, 223
foundational principles of, 23; global DTx commercialization: challenges of,
market, 123, 192–93; guidelines for, 226; channel partnerships as path to,
137–38; identification of pain points, 10–11, 193, 194, 197, 198–99, 242–43,
74; impact on clinical and financial 254–55; common blind spots, 193–94;
outcomes, 87; intellectual property regulatory requirements, 129–30;
and, 98–99; JTBD framework, 58–60; studies of, 195n8
local context, 236; m easures of perfor- DTx pricing: considerations for selecting,
mance of, 137; outcomes of, 27–28; 240–41, 247; DTx as a service approach,
patient-focused nature of, 35, 88, 256; 242; franchise model, 242; geog raphical
potential of, xi, 4–5, 10, 255–56; as considerations for, 242; hybrid models,
prescription-based products, 35, 254; 247; importance of real-world evidence,
private-and public-sector financing 234–35, 245–46; one-off payment
of, 247; problem-solving capabilities, model, 240; private payers, 236–37;
39–40; proliferation of, 192–93; public public payers, 233–36; return-on-
payers, 232–36; real-world evidence investment-based model, 235, 237,
(RWE) of effectiveness of, 30, 31, 246; revenue sharing with distribution
234–35, 245, 255; regulation of, 137, channels, 242; subscription model, 240,
138, 148, 149–50, 194, 232–33, 234, 246; users’ willingness to pay and,
245; regulatory approval, 31–35, 130, 237–39; value-based model, 231, 235,
148, 251; risk-based classification of, 237, 243–46
148–49; roadmap to scalability of, 9; Dutch health care system, 79–80
role of qualitative research in develop-
ment of, 136; stakeholder ecosystem, economic evaluation studies, 129
46, 47, 55; standard of care vs., 133–35; effectiveness vs. efficacy, 137
strategic focus areas, 104; subcategories e-health, 154–55, 161n4, 169
of, 26; test-and-learn approach, 102–4, eHealth Stakeholder Group, 34
110, 111; test for efficacy, 138; users of, electronic health records (EHRs), 6, 8, 165
101, 111n1, 112n7, 156–57; validation emotions, 64
of, 10, 128, 135–36, 137, 150–51; value Employee Health Association, 207
proposition, 104, 134; vs. wellness apps, employers: benefits schemes, 236;
10; work in practice, 24. See also imple- channel partnerships, 206–10; DTx
mentation of DTx solutions innovations and, 207–8, 209; as health
308 Index

care enablers, 206; insurers and, 210; Glooko, 198
value of investment (VOI) framework, Google: AI system, 158; deep learning
207 program, 158; diabetic retinopathy
EndeavorRx, 26, 35, 67, 68, 134 screening system, 158–60; health care
Engine Biosciences, 266 ventures, 221, 223–24; “test and learn”
Entresto, 244 approach in, 102–3
equal value of life years gained (evLYG), Greenhalgh, Trish, 114
129
EU Medical Device Regulation (EU MDR), Hanson Robotics, 263
149–50 Happify, 200
European Medicines Agency, 145 Hardesty, Christopher, 14, 191
European Union: regulatory requirements Harris, Brian, 77, 98
of DTx, 33–34, 123, 149 Harvard Pilgrim Health Care, 245
evidence-based medicine, 37n27 Haven, 223
evidence-based theoretical frameworks HbA1c test, 48, 52
(EBTFs), 270, 272, 273, 275 health care: artificial intelligence in,
Express Scripts, 246 265–66; cost of, 6; evaluation of quality
Exubera inhaler: benefits of, 66; complex- of, 114; expenditure and outcomes-
ity of using of, 66; cost of, 45, 67; emo- based contracts, 244–45; gap between
tional proposition, 67; patient’s per- recommended and provided, 114–15;
spective on, 44–45, 66–67; physician’s national guidelines, 115; new delivery
perspective on, 45; size of, 66 model, 4; pain points in, 64; personal-
ization in, 88; shortage of specialists,
Facebook, 221 81; study of, 114, 115; technology and,
factorial tests, 110 3, 6, 7, 113; value in, 178, 231, 243
Federal Food, Drug, and Cosmetic Act, Healthcare Information and Management
36n17 Systems Society Global Health
5G connectivity, 218, 220 Conference, 125
Food and Drug Administration (FDA): HealthHabit, 224
approval of insulin dosing system, 198; Health Nudges, 186, 188
approval of mobile apps, 24; regulation Hello Heart, 246
of AI, 268; regulation of DTx solutions, Humana Inc., 102–3, 106
191, 194, 234; safety and performance
approvals, 36n17, 82, 149; support of IDentif.AI, 266
decentralized t rials, 145 implementation of DTx solutions: admin-
Ford, Henry, 57 istrators’ perspective, 165; barriers to,
Fujitsu Connected Technologies Limited, 113, 167, 170; data management, 166;
225 data provenance, 163–64; educational
functioning prototypes, 109 and training materials about, 171;
expectations of health care managers,
gamification, 67, 68 163–67; financial and legislative sup-
Germany: Digital Healthcare Act, 37n26, port, 170; importance of right know-
138; Digital Supply Act (DVG), 37n21; how for, 168; integration and interoper-
DTx regulation in, 33 ability f actor, 164, 166; IS approach,
gestational diabetes mellitus (GDM), 70 116, 168–69, 171; key f actors of, 154–56;
Global Access Platform, 205 local context and, 168; NASSS Frame-
Global Epidemic Prevention Platform, 221 work, 170; nurses’ needs and, 160;
Global L
egal Solutions Group, 122 outcomes, 172n4; patient needs and,
Index 309

implementation of DTx solutions (cont.) Kaia Health, 27, 98
159–60; patients’ perspective, 165; Kaissis, Georgios, 122, 123
physicians’ perspective, 159, 160, 161, Kasparov, Garry, 262
165; as a p
rocess, 167, 171; rapid cycle key assumptions checklist, 95
testing approach, 168; regulations and, Kim, W. Chan, 64, 69
166–67; reliable network connection Klinify, 204
for, 159; replicability of, 168; risk con- knowledge, 2
cerns, 166; security and privacy, 164–65, Kodak, 221–22
166; stakeholders, 170; strategic ap- Koh, Dorothea, 180
proach to, 167; structured and phased KT Corp, 218, 220–21
approach, 168; systematic review of, Kumar, Raj, 12, 43
169, 171; technology and, 165, 170; Kurma Partners, 91
traps of, 167–69
implementation science (IS) approach, Lam, Carolyn, 7, 113
115–16, 168–69, 171 Lee, Charlotte, 175, 182, 183
Index of Multiple Deprivation (IMD), LG, 218
75n13 life science organizations, 199–202,
India: rate of gestational diabetes 203–4
mellitus in, 70 l imited memory AI, 262–63
Inevitable Ventures, vii liver transplant patients, 141
ingestible sensors, 22, 36n3, 193 Livongo, 94, 186, 187–88, 207, 209–10,
inner setting, 155 217
Innovator’s Guide to Growth, The, 109 low back pain, 1, 2–3
insomnia, 175, 178, 215–16
Institute for Digital Medicine (WisDM), 6 machine learning (ML), 263, 264, 265,
institutional review board (IRB), 129, 268
142 malaria, 133
insulin pump, 70–71, 72, 75n26 Martucci, Eddie, 14–15, 249, 252, 253
insurance companies: customer experi- mattering theory, 273, 274–75
ence, 210, 211, 212; DTx solutions for, Mauborgne, Renée, 64, 69
211, 212; employers and, 210; partner- McCann, Corey M., 29, 191
ships, 210–13; value-added services, McCarthy, Owen, 77, 78
210–11, 212 McGlynn, Elizabeth A., 114
intellectual property (IP), 98–99, 122 McKesson, 203
International Medical Device Regulators MedCity, 138
Forum, 32, 148 medical devices, 202–3
International Society for Pharmacoeco- medical technology innovation, 162–63
nomics and Outcomes Research MedRhythms, 27, 77, 78, 83, 98, 192
(ISPOR), 130 Medtronic, 70, 71, 72–73, 94
Internet of T
hings (IoT), 220 Megginson, Leon C., 101
investors, 92, 95, 97–98 memory loss, 82
mental health: digitally delivered
Janssen Research & Development, 146 interventions for, 191–92, 234–35,
jobs-to-be-done (JTBD) framework, 253–54, 270–75
57–60, 63 Mercer, 206–7
JOGO, 241–42 mHealth, 156, 161, 161n4
joining data consortia, 118 midwives, 81
JPMorgan Chase, 223 Mildon, Robyn, 13, 154, 162, 173
310 Index

milestone planning chart, 95 Organisation for Economic Co-operation
mindline.sg, 234 and Development, 206, 238
mobile health tools, 2, 24, 142, 169 Osman-R ani, Azran, 74
multiarm trials, 134 Otsuka Pharmaceut ical, 193, 195n8, 200,
multimorbidity, 79–80 201, 204
multiple sclerosis (MS), 78, 87 outcomes-based deals, 244
multivariate tests, 110 outer setting, 155
musculoskeletal disorders, 98 overhead costs, 93
music therapy, 77–78
pain management, 83–84
Naluri, 202, 246 pain points, 64, 66, 74
National Cancer Screening Registry painted door test, 107–8
(Australia), 218 Parkinson’s disease, 77, 78, 82
National Health Service (NHS) (UK), 218 PathAI, 265
National Institute for Health and Care Patient Centered Outcomes Research
Excellence (NICE), 137 Institute, 182
National University Health System patient-centered solutions, 157
(NUHS) (Singapore), 180 patient-physician relationships, 177–78,
net monetary benefit (NMB), positive, 178 180, 182–84, 253–54
Neupro, 133 patients: behavior mechanisms used to
neural network, 265 nudge, 186, 187–88; DTx potential
neurological diseases, 82–83, 193 for, 7, 56, 179–82, 251–52, 256; first
neurologic m usic therapy, 77 consultation with a specialist, 47; peer
Ngiam, Kee Yuan, 13–14, 119, 163 support, 183–84; teleX organizations
N-of-1 (personalized) clinical t rials: and, 214
CONSORT extension for, 142–43; PatientsLikeMe (online community), 183
development of, 86; digital therapeu- Pear Therapeutics, 27, 28n7, 33, 200,
tics for, 88, 151; implementation prob 215–16, 230n19
lems, 29; improvements in data analyt- Penn Medicine, 186, 188
ics, 139–40; intervention sequence, personalized medicine, 7, 86–87
139; IRB review process, 142; as path- Pfizer, 44, 45
way toward individualized treatments, pharmaceut ical companies: challenges of,
143; patient-centric design of, 87, 139; 199–200; data-sharing agreements,
vs. randomized clinical trials (RCTs), 30 201–2; drug-plus programs, 201; DTx
nonadoption, abandonment, spread, partnerships, 82, 193, 200–202, 204–5,
scale-up, and sustainability (NASSS) 226, 254–55
framework, 170 phar maceut ical distributors, 203–4
North East London National Health physicians: caring for patients with
System Foundation Trust (UK), 235 diabetes, 50–51, 52; DTx solutions
Novartis, 244 and, 7, 56, 174, 175–76
nurses, 81 Pichai, Sundar, 123
pilot studies, 113–14
Omada digital DPP program, 213 Ping An Good Doctor, 215, 222
Omada Health, 84 plan-do-study-act (PDSA) method, 168,
Oon, Matt, 15, 89, 270, 272, 273–74 169
OpenMined, 122, 125 prescription medications, 238
operational pivots, 109 price elasticity analysis, 239
ophthalmologists, 81 Privacy Project, The, 123
Index 311

private payers, 232, 236–38 SG&A costs, 93
problem identification, 41–42, 74 Shannon, Jennifer, 128
pro forma operations specs, 95 Shokri, Reza, 124
Propeller Health, 202 Shull, Jessica, 11
Proposed Guiding Principles for Reimburse- Sibel Health, 220–21
ment of Digital Health Products and Sidekick, 200
Solutions, 33 Singapore: AI development in, 268; diabe-
prostate cancer, 86 tes cases in, 48; digitally delivered
Proteus, 192–94, 195n8, 201, 204 mental wellness services, 234–35; DTx
psychotherapy, 84 platform regulations, 32–33, 150; Health
public payers, 232, 233–34, 235–36 Sciences Authority (HSA), 150
Singapore National University Health
quality-adjusted life year (QALY), 129, 178 System (NUHS), 163, 172n1
quiz responses, 106, 107 Siri, 260
SK Telecom, 218
Raku-R aku smartphones, 225 sleep apnea, 202
randomized controlled t rials (RCTs), 29, sleep improvement program, 215–16
30, 98, 128, 133, 138 Sleepio (DTx solution), 134, 175, 178,
rapid cycle test, 168, 169 183
reach52, 216 small data, 87, 141, 152n11
reactive AI, 262 smoking cessation program, 110
real-world data (RWD), 30, 114, 116, 132, software as a medical device (SaMD), 32,
252 148, 202, 268
real-world evidence (RWE), 30, 31, 128, software as a service (SaaS), 239, 247n4
151, 234–35, 245–46, 268 Somryst, 215–16
real-world outcomes, 128 Souillard, Fabien, 12, 43
relaxation, 2 stakeholders, 41, 57, 64, 65, 73
remote decentralized clinical t rials standard of care, 133–34, 152n6; DTx
(RDCTs), 143–44 solution vs., 133–35
Remus, Alexandria, 259 startups: fundraising, 99; investors of,
Repatha, 245 96–98, 99
reSET products, 29, 33 stroke therapy, 82, 83, 183
ResMed, 202 strong AI, 259, 260, 261–62
resource-intensive diseases, 81 super AI, 259, 260, 262
return on investment (ROI): vs. value of supplements, 31–32, 36n17
investment (VOI), 207 symptoms monitoring solutions, 36n3
reverse income statement, 92, 93, 94, 96
RhythmAnalytics, 33 Tadeo, Xavier, 128
Rhythmic Auditory Stimulation (RAS), 78 Takeda, 203
task-technology fit model, 155
SaaS approach, 246 Tata Consultancy S ervices, 146
Savonix, 224–25 technological innovations, 5, 40–41, 114
schizophrenia, 82, 193, 272 technological solution’s ecosystem,
self-aware AI, 263 45–46
serv iceable available market (SAM), 91, technology acceptance model, 155
92, 99n1 technology implementation: contextual
serv iceable obtainable market (SOM), 91, factors, 157–59; factors affecting,
92, 99n1 154–55; usability and, 156–57
312 Index

Teladoc, 94, 215, 217 United States: diabetes therapy nonadher-
telcos (telecommunications companies): ence, 53; DTx regulations, 34; health
B2C and B2B2C business models, 219; care expenditures, 244
data privacy and security regulations, UpScript Health, 215–16
218; health care services, 218, 220; Us2.ai medtech company, 7
partnerships, 217–21, 226; reputational usability: as factor affecting new
risk, 219; subscription-based models, technology, 156–57
219; technical capacities, 218 utility levers, 64, 65, 66, 68
telehealth (teleX): basic models, 214;
cross-selling potential, 217; delivery of value: concept of, 193–94
health care, 214–15; DTx partnerships, value-added services (VAS), 210, 211–12
214–17; expansion of, 7, 214–15; laws value-based contracting schemes, 247
and policies, 214; patient-provider value-based health care: development
interaction, 214; privacy concerns, 214; of, 231, 232; DTx pricing and, 243;
user satisfaction, 215 patients’ perspective on, 232, 243–44;
Telstra, 218 providers’ perspective on, 232–33
“test and learn” culture, 101, 102–3, value of investment (VOI): vs. return on
110 investment (ROI), 207
theory of mind AI, 263 video games, 134
therapy nonadherence, 53–54 Voluntis, 28
Time BioVentures, vii
tobacco users: engagement with digital waitlist-based control groups, 134, 135
tools, 183 Wallach, D. A., vii
total addressable market (TAM), 91, 92, weak AI, 259, 260–61
99n1 Welldoc Inc., 24, 28, 241
transaction tests, 109 wellness and support, 21, 22
traumatic brain injury, 82 wellness apps, 10
treatment: dosing adjustments, 86; Wellthy Therapeutics, 202
personalized approaches, 85–86; side World Health Organization, 39, 53
effects, 86; use of artificial intelligence, Wysa, 235
86, 140–41
Yeo, Geck Hong, 89, 270, 272–73, 274
UCB Biopharma, 133 Yeoh, Ester, 11–12, 43, 47, 51, 52, 55
unified theory of a cceptance, 155
United Kingdom: DTx regulations, 33–34 Zuellig Pharma, 203–4
Index 313

About the Authors
Darren Gabriel Leow
Dean Ho
Provost’s Chair Professor, Director of the Institute for Digital Medicine (WisDM),
Director of the N.1 Institute for Health, and Head of the Department of Biomedical
Engineering at the National University of Singapore
Dean Ho and collaborators successfully developed and validated CURATE.
AI, a powerf ul artificial intelligence platform that personalizes h uman
treatment for a broad spectrum of indications ranging from oncology to
digital therapeutics and infectious diseases, among o thers.
Ho is an elected fellow of the National Academy of Inventors. He is also
a fellow of the American Association for the Advancement of Science, the
American Institute for Medical and Biological Engineering, and the Royal
Society of Chemistry. He was named to the Healthcare Information and
Management Systems Society’s Future50 Class of 2021 for his internation-
ally recognized contributions to digital health and is a subgroup lead in the
World Health Organization’s Working Group on Regulatory Considerations
for AI in Health.
Ho has appeared on the National Geographic Channel’s Known Universe
and Channel News Asia’s The Hidden Layer: Healthcare Trailblazers. His
discoveries have been featured on CNN and NPR and in the Economist,
Forbes, and the Washington Post, among other international news outlets.
He has also served as the president of the board of directors of the Society

for Laboratory Automation and Screening, a global drug development
organization comprised of senior executives from the pharmaceutical and
medical device sectors as well as academic visionaries.
Esmond Lim
Yoann Sapanel
Head of Health Innovation, the Institute for Digital Medicine (WisDM) at the
Yong Loo Lin School of Medicine, National University of Singapore
Yoann Sapanel, passionate about bridging the efficacy-effectiveness gap
for digital therapeutics, is an expert in medical technology development
driven by payer perspectives.
At WisDM, Sapanel aims to foster and accelerate cross-industry,
public–private collaborations to validate, implement, and scale pioneering
pharmacological and digital interventional technologies into the clinic. As
a member of the Singapore Health District initiative, he designs, evaluates,
and drives the implementation of technological solutions within the
community to support resident’s health through life stages.
Prior to that, Sapanel was the head of Health Partnerships at MetLife,
focusing on the development of new products and serv ices grounded in
technology and data for cancer, dementia, and cardiovascular diseases. In
that role, he led the implementation of the world’s first mobile, clinically
valid, neurocognitive assessment test and mobile app for improving brain
health in Japan.
Earlier, Sapanel led the Asia Pacific Center of Excellence for Business
Model Innovation at Medtronic, where he designed, executed, and scaled
the company’s first specialized clinic for the management of heart failure
patients. As program director for Medtronic Asia’s Value-Based Health
Care Council, he was instrumental in advancing the adoption of outcome-

based models to ensure better patients outcomes, reduce costs, and deliver
value across the region.
As consultant at Blue Ocean Strategy and PwC, Sapanel also conducted
strategy formulation projects with leading pharmaceutical companies.
Nicholas Brocklebank
Agata Blasiak
Research Assistant Professor and Head of Digital Health Innovation at the
Institute for Digital Medicine (WisDM) and the N.1 Institute for Health at the
National University Singapore
Agata Blasiak is a developer and implementer of digital health technolo-
gies. She has codeveloped several digital platforms for decentralized health
care that are in the process of being clinically validated and has collabo-
rated with leading health startup innovators. Beyond technical develop-
ment, Blasiak’s research areas of interest are behavioral and societal
aspects of digital health solutions as catalysts for the holistic translation of
inventions into innovations that have a positive impact on patients.
At WisDM and N.1, Blasiak’s work has focused on leveraging CURATE.AI
for personalized dosing in oncology. Her work has also harnessed IDentif.
AI—an AI platform for optimizing combination therapies for infectious
diseases—to rapidly pinpoint actionable drug combinations against
SARS-CoV-2 and other infectious diseases. She has received multiple
awards for her work, including MIT Technology Review recognition as a
member of the Innovators Under 35 Asia Pacific 2021.
Before venturing into the digital health space, Blasiak was a neuroengi-
neer researcher and focused her efforts on molecular neuroengineering and
developing neuroimplants. She holds a BScEng in biotechnology from
Warsaw University of Technology, Poland, and a PhD in bionano interac-
tions from University College Dublin, Ireland.

Ebook About Biology Research AAA

Uploaded by

Copyright:

Available Formats

Ebook About Biology Research AAA

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ebook About Biology Research AAA

Uploaded by

Copyright:

Available Formats

MED I CINE W I T H O U T MED S

Ho_Meds_int_3pgs.indd 1 14/06/23 12:45 PM

JOHNS HOPKINS UNIVERSIT Y PRESS | Baltimore

Ho_Meds_int_3pgs.indd 3 14/06/23 12:45 PM

Johns Hopkins University Press

Library of Congress Cataloging-in-Publication Data is available.

ISBN 978-1-4214-4703-2 (hardcover)

Ho_Meds_int_3pgs.indd 4 14/06/23 12:45 PM

Foreword, by D. A. Wallach vii

Introduction. The Context 1

PART I. WHAT ARE DIGITAL THER APEUTICS? 19

PART II. THE RIGHT TOOL FOR THE RIGHT PROB­L EM 39

PART III. FROM PI­L OT TO PRODUCTION 113

Ho_Meds_int_3pgs.indd 5 14/06/23 12:45 PM

Conclusion. The F­ uture of DTx 249

Ho_Meds_int_3pgs.indd 6 14/06/23 12:45 PM

Through the Lens of a Consumer and Tech Investor

Ho_Meds_int_3pgs.indd 7 14/06/23 12:45 PM

Building Ventures to Reinvent Health Care

Ho_Meds_int_3pgs.indd 8 14/06/23 12:45 PM

Ho_Meds_int_3pgs.indd 9 14/06/23 12:45 PM

Ho_Meds_int_3pgs.indd 11 14/06/23 12:45 PM

Ho_Meds_int_3pgs.indd 13 14/06/23 12:45 PM

Ho_Meds_int_3pgs.indd 1 14/06/23 12:45 PM

2 MEDICINE WITHOUT MEDS

Ho_Meds_int_3pgs.indd 2 14/06/23 12:45 PM

• Cardiovascular diseases: the leading cause of death globally,

The list goes on b

Ho_Meds_int_3pgs.indd 3 14/06/23 12:45 PM

The consequences of an intensifying chronic disease epidemic are

Existing Solutions in Health Care Are Not ­Adopted

4 MEDICINE WITHOUT MEDS

Ho_Meds_int_3pgs.indd 4 14/06/23 12:45 PM

• How ­w ill DTx be implemented?

Technology Alone Cannot Transform Health Care

Ho_Meds_int_3pgs.indd 5 14/06/23 12:45 PM

interactional ele­ment is what opens doors and pushes new bound­

6 MEDICINE WITHOUT MEDS

Ho_Meds_int_3pgs.indd 6 14/06/23 12:45 PM

Ho_Meds_int_3pgs.indd 7 14/06/23 12:45 PM

Who Is This Book For?

8 MEDICINE WITHOUT MEDS

Ho_Meds_int_3pgs.indd 8 14/06/23 12:45 PM

Realizing Medicine 4.0

Ho_Meds_int_3pgs.indd 9 14/06/23 12:45 PM

10 MEDICINE WITHOUT MEDS

Ho_Meds_int_3pgs.indd 10 14/06/23 12:45 PM

Introducing the Key Contributors

Dr. Jessica SHULL

Dr. Ester YEOH

Ho_Meds_int_3pgs.indd 11 14/06/23 12:45 PM

Dr. Raj KUMAR

12 MEDICINE WITHOUT MEDS

Ho_Meds_int_3pgs.indd 12 14/06/23 12:45 PM

Dr. Kee Yuan NGIAM

Ho_Meds_int_3pgs.indd 13 14/06/23 12:45 PM

Dr. Eddie MARTUCCI

14 MEDICINE WITHOUT MEDS

Ho_Meds_int_3pgs.indd 14 14/06/23 12:45 PM

PART II. THE RIGHT TOOL FOR THE RIGHT PROBL EM 39

PART III. FROM PIL OT TO PRODUCTION 113

Conclusion. The F uture of DTx 249

Existing Solutions in Health Care Are Not Adopted

• How w ill DTx be implemented?

interactional element is what opens doors and pushes new bound

As an independent category within digital health, the appeal of dig-

Similarly, DTx must also adhere to ten foundational principles:4

• DTx used as standalone treatments (a.k.a. drug-replacement

DTx Differential Advantage #2: Real-World Evidence Generation

Real-world evidence is of greatest importance since clinical trials are only

Due to the dynamic nature of the DTx development process in re-

• In Singapore: Currently DTx fall u nder the definition of

hese regulatory advances reflect the multifaceted transformation of

❒ What is your DTx organization’s primary purpose and objec-

40 The Right Tool for the Right Problem