Computers & Security 133 (2023) 103404

Contents lists available at ScienceDirect

Computers & Security

journal homepage: www.elsevier.com/locate/cose

Forecasting cyber security threats landscape and associated technical trends

in telehealth using Bidirectional Encoder Representations from
Transformers (BERT)
Usharani Hareesh Govindarajan a, b, *, Dhiraj Kumar Singh c, Hardik A. Gohel d
a
Business School, University of Shanghai for Science and Technology, Shanghai, China
b
School of Intelligent Emergency Management, University of Shanghai for Science and Technology, Shanghai, China
c
Edifecs, Inc. Bellevue, Washington, United States of America
d
Applied AI Research Laboratory, Department of Computer Information Sciences, University of Houston–Victoria, Texas, United States of America

A R T I C L E I N F O A B S T R A C T

Keywords: Purpose: Patient enablement and digital patient records have become crucial for the new order in healthcare
Telemedicine delivery, as patients start to take an increased hands-on approach in their healthcare curation. A broader un­
Cyber security derstanding of the efficacy and security of medical data in specific configurations of clinical interventions, ap­
Insecure mobile interfaces
plications, technological structures, logistics, and contextual situations is an urgent need. This research,
Patent analysis
Emerging technical trends
therefore, analyzes 12,582 global patent grants to visualize the technical trend across seven major security
Financially motivated attack threats identified by Kim et al., in their research titled, “Risk management-based security evaluation model for
Data motivated attack telemedicine systems”. The paper further investigates the adoption framework focusing on the stakeholders.
Click fraud Methodology: The methodology used presents a systematic visualization across a large corpus of global patent
Adware grants from over 40 key patent offices to better understand the cybersecurity technology landscape and emerging
Proxy malware trends for Telehealth. This visualization is enabled by a framework for intelligent collaborative patent mining
Banking trojans which uses Bidirectional Encoder Representations from Transformers (BERT) for topic generation embedding
Charge-ware
text data to high dimensional mathematical space for contextual information mining within the patent grants
Spyware
space.
Findings: The study identifies and categorizes across the seven major threat groups a significant amount of
technical building blocks focused on the network and devices, however, it is found that development accounting
for the end user (patient and associates) aspects such as diversity, education, emotions, etc. are yet to received
attention a consideration during technical component development. The summarized results with a focus on
human aspects among others will serve as a guide for using telemedicine to triage patients to the appropriate
level and source of care.

1. Healthcare and electronic health/medical records in the world but has also changed the format of medical records and, in
turn, healthcare (Dash et al., 2019). Recent improvements in the med­
The term medical record has gained a lot of attention in recent years. ical records industry aim to further professionalize systems to further
This is because patient medical information should include health and streamline workflows, increase productivity, and improve
lifestyle information, not just incidental medical encounters. Tradi­ doctor-patient interactions (Sheingold et al., 2014). The widespread
tionally, medical records were written on paper and kept in folders that adoption of EHR systems has long been hailed as transformative for the
were sectioned based on the nature of the notes, and only one copy was healthcare industry. 1–3 The HITECH Act, passed in 2009, authorized
available. New computer technology developed in the 1960s and 1970s more than $30 billion in spending to increase EHR adoption. Since then,
laid the foundation for the development of the Electronic Health Record the number of hospitals with a basic EHR system has increased
(EHR). The use of electronic medical records has not only made patient dramatically, rising from 15.6% to 75.5% between 2010 and 2014
medical information easier to read and available from nearly anywhere (Yuan et al., 2019). Electronic medical records (EMRs) are widely used

* Corresponding author.
E-mail address: [email protected] (U.H. Govindarajan).

https://doi.org/10.1016/j.cose.2023.103404
Received 17 January 2023; Received in revised form 23 June 2023; Accepted 23 July 2023
Available online 30 July 2023
0167-4048/© 2023 Elsevier Ltd. All rights reserved.
U.H. Govindarajan et al. Computers & Security 133 (2023) 103404

in E-health applications around the world because they help to ensure publication with patent portfolios followed by a conclusion and future
safe and high-quality healthcare services by performing a variety of scope in Section 7. This research provides knowledge transparency to­
functions such as documentation, medication management, practice ward secure Telemedicine infrastructure by scrutinizing the threat
management, and communication (Buvik et al., 2016). EMRs manage a landscape and mining-associated technical trends. The presented
wide range of digitally organized patient data, such as demographic knowledge transparency affords stakeholders (industrial, and academic)
information, medical history, radiology, and laboratory results. Between the to acquire the awareness of secure telemedicine services.
2011 and 2021, the average adoption rate of the sampled hospitals in
China increased from 24.2% to 92.3%, compared to a 16.6% to 96.7% 2. Healthcare infrastructure and EHR
increase in US hospitals. The annual average adoption rates in Chinese
and American hospitals were 6.81% and 8.01%, respectively. However, The healthcare infrastructure consists of majorly 4 pillars in play as
the annual average number of hospitals adopting EHRs in China was shown in Fig. 1. Patients are the center of this infrastructure. Healthcare
1667 versus 621 in the US, indicating that the former may necessitate provider consists of hospitals, imaging facilities and all the different
more effort. The adoption rates of hospital EHRs in China and the United points of contact for a patient to get treatment. Insurance companies
States have both increased significantly in the past 10 years (NIH). At insure the patients. Healthcare operation management companies help
present almost every hospital in US and China are equipped with EHRs. providers and insurance companies with inefficient and healthcare-
This faster adoption outcome may have been a benefit of top-level related operations. These companies mostly deal with the technological,
design and government-led policies, particularly the inclusion of EHR infrastructural, and analytical side of operation management. Finally,
adoption as an important indicator for performance evaluation and the the government body is the main pillar that enforces healthcare data
appointment of public hospitals. security among the other three pillars.
Despite widespread adoption, two major challenges remain. The first All these four participants are responsible for patient data manage­
difficulty is, security because of the unique nature of doctor-patient ment. There is a huge amount of patient data exchange between the four
confidentiality, concerns about electronic data and privacy have sha­ and since all of them are working on patient.
ped both public policy and private software development. HIPAA reg­ Furthermore, healthcare firms are expected to bear outrageously
ulations, for example, were created to address the security of patient expensive compliance costs as a result of the legislation, potentially
medical records. The second issue is the EHR’s interoperability. There reducing healthcare firms’ profitability and driving them to avoid
are still challenges in this area, and both the public and private sectors compliance (Khansa et al., 2012). In comparison to other data in­
are working to improve the security of medical records at all access and dustries, the healthcare industry has been hit the hardest. Furthermore,
transmission. To better understand these challenges towards plausible according to the IBM Cost of a Data Breach Report, 2019, the cost of a
solutions, this research conducted a review to understand the develop­ breached record in the healthcare industry increased by 19.4 percent,
ment in information security to tackle these challenges faced by the the greatest increase in this period. According to Seh et al., the health­
telemedicine ecosystem. A structured search of bibliometric and patent care business was swamped by hackers from 2015 to 2019, with 90.49
databases for 6 years between 2016 and 2022, using appropriate fre­ percent of health records being compromised within this period only
quency count optimized keyword optimization. A collection of 12,582 (Seh et al., 2020). According to the Kaspersky lab report 2019, ran­
global patent documents were analyzed and reviewed. Cross-Validation somware and zero-day assaults have surged significantly. Hacking/IT
is conducted along with subject matter expert verification for the ac­ incidents are now the most common causes of healthcare data breaches,
curacy of information. This paper is organized into six sections. Sections according to the HIPPA journal Healthcare Data Breach Statistics
2 and 3, present essential background knowledge required to understand (collected till December 2021). unauthorised access/disclosure in­
the advent of telemedicine and the security concerns within the infra­ cidents are also common. According to the publication, data breaches
structure. Section 4 presents the data gathering and exploratory data have increased from one per day in 2018 to 1.95 per day in 2021 (HIPPA
analysis. Section 5 explains the systematic methodology toward search, journal). Although current e-healthcare providers claim that patients
filtering, and knowledge extraction. Section 6 is a consolidation of can only access their electronic medical records; in the recent past
method validation and key findings from extracted knowledge several healthcare data intrusions and breaches have been reported in

Fig. 1. Health Infrastructure Nexus.

2
U.H. Govindarajan et al. Computers & Security 133 (2023) 103404

the USA and various European countries (Yaqoob et al., 2017); wherein in numerous areas have resulted in growing concerns about information
medical data of millions of patients have been breached. Thus, there is a security. As a result, the adaptation and application of IT in healthcare
need to continuously upgrade the security standards for sensitive health should be closely monitored to ensure access control, proper confiden­
data. tiality, authorization, and authentication procedures. The problem with
confidentiality, poor security, and reliability provided by caregivers
3. Telemedicine and the increasing data security concerns should be addressed (Luciano et al., 2020). Edimara and her team also
emphasize that another challenge is the use of various types of tech­
Telemedicine is a shorthand for remote electronic patient encounters nology to access health records, such as records retrieved via mobile
using telecommunications technologies that provide health information devices. One of the most significant challenges in electronic and ubiq­
and healthcare services. Mobile health is a recent notion that has uitous healthcare is the security of IoT devices (including personal de­
emerged as a result of the expansion and development of the smartphone vices) and the underlying healthcare information infrastructure
and internet of things (IoT) industries. Personal digital assistants, (Soltanisehat et al., 2020). Moustafa and his team highlight critical IoT
wireless patient monitoring devices, smartphones, and tablet computers security challenges that the healthcare industry will face (Mamdouh
are examples of mobile communication devices that support health care. et al., 2021). Furthermore, mobile phone applications will provide ac­
Mobile applications (apps) and, in some cases, associated mobile devices cess to additional patient information, which will aid businesses in
and sensors are the facilitators and drivers of mobile health systems. profiling patients for monetary gain violating their privacy (Ribeir­
This has enabled clinical services to leverage telecommunication, im­ o-Navarrete et al., 2021). To achieve the resilience levels required for
aging, and information technology to provide more efficient healthcare. protracted digital transformation, healthcare organizations must
Telemedicine can make healthcare cost-effective and convenient by strengthen their adaptive, transformative, and absorptive capabilities in
reducing the patient load in hospitals because of technical healthcare the cyber security context (Garcia-Perez et al., 2023). To understand the
outreach. Moreover, telemedicine bridges the user experience gap be­ security flaws, it is important to understand the PHI data production to
tween the healthcare provider and the patient giving patients better analytics flow described in Fig. 2.
access to their healthcare provider and reducing absenteeism. Fig. 2 shows the PHI data flow in telemedicine services between the 3
With the rise of relevant technologies and IoT, the healthcare in­ major players. Patients are the PHI data producers. Produced data is
dustry is redeploying toward healthcare 4.0, the core concepts of which then collected from various sources like medical testing facilities,
include cloud-based centralized context-aware services. Furthermore, as communication with the patients, patient’s sensors, etc. collected data is
big data analytics becomes more prevalent, the trend in healthcare further processed so that it can be saved in electronic form efficiently
services is shifting toward individually tailored medical patient care and is query-able. PHI Data flow can be divided into two parts. The first
(Geum and Kim, 2020). The ongoing Covid pandemic and the concept of flow is between the patient and the provider and the second flow is
retaining distance have accelerated breakthroughs in the realm of tele­ between the provider and a third-party organization (Healthcare IT and
health (Kichloo et al., 2020). One of the most pertinent responses to this Analytics management). The second form of data sharing existed before
pandemic in health care has been the widespread deployment of tele­ telemedicine and has better orchestration for data security and privacy
medicine as a surrogate for in-person care around the world. Telemed­ but still faces a lot of cyber attacks. On the other hand, data sharing
icine advancements assist in fighting this pandemic by focusing on three between patients and providers is relatively new and contains a lot of
primary goals. The first goal is to care for the surge of infected patients threat points as pointed out by Kim et al. work. The seven major threats
who require isolation and intensive care. Second, it allows focus on outlined by Kim et al., when it comes to data flow from the patient to the
routine patient health condition monitoring, and finally, it protects provider are presented in a diagram in Fig. 3.
clinicians and patients from infections through tracking (Monaghesh Threat #1: Demographics: Telemedicine patients are typically
and Hajizadeh, 2020). Digital patient empowerment and remote area residents or elderly people and most of them have never
patient-enablement health have become crucial in this era of healthcare, received cybersecurity training and express little curiosity in doing so.
as patients are urged to take increasing responsibility for their treat­ As a result, their use of telemedicine terminals exposes them to security
ment, either alone or in collaboration with their healthcare (Shokri and risks such as device misuse, weak passwords, device loss, social engi­
Lighthall, 2020). Thus, a great majority of medical practitioners have neering, phishing, and so on. According to the United States Department
begun adopting telemedicine with little or no planning or preparation of Justice, social engineering attacks are among the world’s most serious
(Wosik et al., 2020). One major benefit of the telemedicine architecture threats. According to the cyber security analyst firm Cyence, the United
is the ability to provide timely and tailored clinical decisions based on States was indeed the nation aimed primarily by the majority of attacks
the analysis of patients’ data using big data which further helps in involving social engineering in 2016 and incurred the highest intrusion
proper care and medications (Cegarra-Sánchez et al., 2020). cost, accompanied by Germany and Japan. In the United States, the cost
The adoption of telemedicine and interactive software frameworks estimate of these attacks was $121.22 billion. These businesses lost over
involves the digital collection, transfer, and use of sensitive medical $2.3 billion. Furthermore, according to recent surveys and studies, 84%
information between patients and medical practitioners, which may of cyber-attacks are carried out by social engineers with a high rate of
lead to the unintended disclosure of sensitive personal data (Bokolo, success (Salahdine et al., 2019).
2021). There are guidelines for telemedicine also in PHI data security Threat #2: Telemedicine devices: A telemedicine terminal is either
rules, but compliance is not enforced strictly. The healthcare domain based on an embedded-type real-time operating system (RTOS) or a
lacks proper and detailed threat modeling examples which is a necessity general-purpose operating system (GPOS). RTOS-based devices are
for information security (Sönmez et al., 2022). The risk of patient per­ secure from unauthorised access because they have been structured for
sonal information disclosure in healthcare increases with the transfer of specific functions during the manufacturing and design stages. GPOS-
large volumes of patient treatment information between various treat­ based devices, in contrast, such as smartphones, raise security con­
ment departments and individuals (Rezaei et al., 2021). Baltzan defines cerns due to the presence of third-party apps. Because of these systems’
information security as “the protection of information from accidental data storing and sharing capabilities, as well as the risk of app vulner­
or intentional misuse by persons inside or outside an organization”. abilities, device loss/theft, and plaintext transmission, utilization of
Potential threats to IT infrastructure and security are viruses and telemedicine terminals in such situations expose them to security con­
hackers. The major blockade in the adoption of Telemedicine is infor­ cerns. An attacker can create a malicious app that looks like a tele­
mation security as information in telemedicine frequently travels over medicine provider app and host it on the Play Store and App Store.
the Internet (Baltzan and Phillips, 2014). Furthermore, one practice that requires scrutiny in the medical devel­
The recent adoption and wide-scale implementation of telemedicine oper space is that of developers’ privacy policies and data security

3
U.H. Govindarajan et al. Computers & Security 133 (2023) 103404

Fig. 2. Telemedicine Architecture Data Flow.

Fig. 3. Telemedicine Security Threats.

practices (Galetsi et al., 2022). system primarily via a wireless network. Communication networks in
Threat #3: Home network: The patient’s personal space (office or the home include Wi-Fi, LAN (local area network), NFC (near field
home) telemedicine terminal communicates with the telemedicine communication), Bluetooth, and third/fourth generation/long-term

4
U.H. Govindarajan et al. Computers & Security 133 (2023) 103404

evolution networks. In such cases, home network-based telemedicine of telehealth interactions are doctor-to-patient (D2P) and doctor-to-
service systems are vulnerable to security flaws associated with man-in- doctor (D2D). The monitoring and exchange of medical information
the-middle (MITM) attacks and end-to-end plaintext transmission. The distinguish D2D telemedicine, which necessarily involves a greater level
old wireless security mechanism is based on wired equivalent privacy of information security due to the virtual consultation, which in­
(WEP), which has security weaknesses such as the use of RC4 and corporates the drafting of prescriptions. The telemedicine system may be
CRC32. Wi-Fi passwords can be compromised as a result of the inade­ vulnerable to security risks such as MITM attacks, malicious code,
quate security system, and data can be acquired via Wi-Fi Sniffers telemedicine app forgery/alteration, and illicit access to medical infor­
(Anwar et al., 2019). The police department in Cambridgeshire, En­ mation by circumventing physical security checks in this environment. It
gland, initially reported a surveillance attack in August 2005 using a could also be vulnerable to security risks associated with prescription
Bluetooth device. Bluetooth attacks can be further classified into 7 cat­ changes, device misuse, and information disclosure.
egories as shown in Fig. 4 (Hassan, 2018).
Threat #4: Gateway devices: A gateway serves as an intermediary 4. Emerging trends disclosed across global patent grants
for the telemedicine system and the patient, exposing the system to se­
curity risks such as gateway loss/theft, rogue gateways, and MITM at­ A patent is a transitory monopoly conferred for commercializing a
tacks. Within this category, various sorts of attacks with varying degrees recently invented device. A patent implies that such a device would not
of severity can be addressed. Traffic analysis is a methodology used by be obvious to a skilled practitioner of the relevant technology. Retaining
attackers to determine the communication load, the size of the packets, an up-to-date evaluation of technical trends and organizational portfo­
the source, and destination of the packets sent and received the number lios enhances technical know-how (Trappey et al., 2019). The global
of packets sent and received. The attacker uses a wireless card capable of patent used in this section data has been extracted using the patent
being set to promiscuous mode, in addition to special antennae such as a search and analysis software IncoPat (IncoPat, 2022). The database
Yagi antenna as well as the global positioning mechanism, to carry out provides wide coverage of more than 120 authorities such as the World
this attack (GPS). Besides that, there is a broad array of unrestricted Intellectual Property Organization (WIPO), the United States Patent and
software available, including Kismet, Netstumbler, and others. An Trademark Office (USPTO), and the European Patent Office (EPO). The
attacker attempts to block or prohibit regular network communication section extends to the patent data analysis to extract the knowledge of
by flooding a genuine client with incorrect messages, fake packets, or trends from the databases for a global perspective around all these
duplicate IP/MAC addresses. Gaurav and his colleagues propose a threats for safe mitigation. The patent dataset is queried in two ways.
method for recognizing DDoS attacks and differentiating them from the First, we query for a generic search over all the patents in the tele­
flash crowd for the general public and small entrepreneurs (Gaurav medicine security space. We further take the granularity a step further
et al., 2022). by looking at patents specific to the threats to analyze each threat’s
Threat #5: Internet (public network): A public network is used for patent space. Table 1 shows the search query and the number of patent
communication between the telemedicine system (the Internet) and the grants associated with that query for further research reproducibility.
patient. Because medical, private, and prescriptions, as well as health The general trends in patent filling have observed exponential
information, are communicated via the public Internet, it is critical to growth since 2016. China has filed the maximum number of patents,
establish end-to-end security requirements. The telemedicine system is followed by The United States and South Korea. The general trends are
vulnerable in this setting to security risks such as sniffer, forgery/ understood by reviewing patents through the technical features of their
alteration, and privilege escalation. This threat is resolvable with content using standardized classification systems set by global agencies.
stringent guidelines during the Telemedicine application development The United States Patent and Trademark Office (USPTO) and the Eu­
process. ropean Patent Office (EPO) jointly manage the Cooperative Patent
Threat #6: Telemedicine system: The telemedicine system is located Classification (CPC) system, which is an extension of the International
at the telemedicine service provider’s location. It comprises a compu­ Patent Classification (IPC) system. Each classification term in the CPC
tation system and the software required for remote consultations, and its system consists of a symbol such as "G16H10/60″ that classifies the
users include nursing staff, medical personnel, and system administra­ patent to technologies handling healthcare or patient-related medical
tors. This system is critical because it manages all the data for the pa­ data specifically for electronic patient records (European Patent Office,
tients who receive telehealth consultations. Telemedicine technologies 2021). The analysis of 12,582 grants was retrieved by scouring the
are sometimes used in telemedicine clinics to provide a wireless Incopat database for patents titled or containing the keywords “Tele­
connection between patients’ fitness equipment and computers being medicine”, “telehealth”, “security”, “IT security”, “Mobile device secu­
used for remote consultation. In these circumstances, telemedicine sys­ rity”, “secure communication”, “healthcare”, “health application
tems may be susceptible to threats such as telemedicine app forgery/ security”, “secure mobile communication”, “healthcare provider”,
alteration, malicious code, MITM attacks, and illegitimate access control “secure telemedicine system”, “mobile application security” and “IoT
via physical security screening circumvention. device security” in the title, abstract, and claims section. This data is
Threat #7: Telemedicine service provider: The most common forms analyzed for the years 2016 to 2022. Table 2 presents patent grant

Fig. 4. Bluetooth Device Threats.

5
U.H. Govindarajan et al. Computers & Security 133 (2023) 103404

Table 1 Technology Licensing, LLC which collectively have 67 patents. It is

Patent data: queries and the corresponding patent grants. interesting to note that China and the United States hold the most
Keywords Search query Year limit Patent number of patents in the notified period, this allows both developed and
grants developing countries to follow best suited to their peculiar scenarios.
Telemedicine, telehealth, ((ALL =(Telemedicine) 2016–01–01 12,582 The analysis of the patent dataset in the observation period shows
security, IT security, OR ALL =(Telehealth) to that maximum patents were granted to CPC code H04L. The code H04L
Mobile device security, OR ALL =(Healthcare) 2022–12–31 focuses on the transmission of signals having been supplied in digital
secure communication, OR ALL =(Health form and includes telegraphic communication, data transmission, or
healthcare, health application)OR ALL
application security, =(Phone consultation)
methods or arrangements for monitoring. This is in line with the threat
secure mobile OR ALL =(Health app) space seen in the healthcare telecommunication and data transmission
communication, OR ALL =(Patient)OR domain. The subclass H04W covers communication networks that
healthcare provider, ALL =(healthcare selectively establish one or more wireless communication links between
secure telemedicine provider) OR ALL
users and network equipment or the desired number of users for the goal
system, mobile =(Doctor)) AND (ALL
application security, IoT =(security)OR ALL =(IT of transmitting information over these wireless transmission links.
device security security) OR ALL Subclass H04B focuses on the transmission of information-carrying
=(Mobile device signals, regardless of the nature of the information, and includes
security)OR ALL monitoring and testing arrangements, as well as noise and interference
suppression and limitation. Subclass G16H covers aspects of computer,
=(secure
communication)OR ALL
=(Phone security) OR information, or communication science that intersect with medical or
ALL =(Mobile security) healthcare science, with a clear emphasis on data processing systems or
OR ALL =(secure digital computing or digital computing or methods that are specifically
telemedicine system)
adapted for medical or healthcare science. The code G06F focuses on
OR ALL =(IOT device
security)) AND (PD= digital data handling and processing. Fig. 5 represents the yearly dis­
[20,160,101 TO tribution across the CPC main and the highlighted disruptions the
20,221,231])) pandemic has brought on new publications, during the period.
Research and investment were on an uptrend till 2019, Covid-19
disruption made a little change in the slope but again the publication
Table 2 trends show increased corporate investment into new research. Table 3
Applicant publication trends by top CPC. shows the top 5 CPC (main) codes, their ranks concerning the number of
Top Assignees Top CPC Codes
patents, and patent counts along with the corresponding CPC groups.
Assignee Patent CPC (4 Patent The CPC main for H04L includes the Transmission of digital informa­
Count characters) Count tion. Amongst these, the CPC group H04L63 has the highest patent count
Samsung Electronics co., ltd. 241 H04L 3312 of 1112 and deals with network communication protocols for network
ZOLL Medical Corporation 52 G16H 2729 security. Similarly, H04L67 has the second-highest patents at 997 and
International Business Machines 24 H04W 2658 deals with Network arrangements or protocols for supporting network
Corporation services or applications. H04L5 has 690 patents that deal with ar­
Capital One Service LLC 19 G06F 2381
Medtronic Inc. 16 A61B 1993
rangements affording multiple uses of the transmission path. H04L9
Allied Security Trust, Siemens 16 G06Q 1680 focuses on arrangements for secret or secure communications, Network
Healthcare GMBH, Siemens Medical security protocols. The second most common patent filing category of
Solutions USA, Inc. G16H which focuses on healthcare informatics has 2729 patents, it
Elira Inc. 14 A61M 1048
highlights the research’s importance in the healthcare industry, and the
Microsoft Technology Licensing, LLC 13 H04B 511
Titan Health & Security Technologies, 12 A61N 388 impact of patents on understanding the security around healthcare data
Inc.: WTI Fund X, Inc.: Venture will benefit secure transmission and management of data. G16H40 has
Lending & Leasing IX, Inc. 1366 patents and includes ICT specially adapted for the management or
Koninklijke Philips N.V. 12 G06K 381 administration of healthcare resources or facilities. G16H10 with 1250
patents is an important sub-class that focuses on ICT specially adapted
trends of various applicants helpful in categorizing the domain tech­ for the handling or processing of patient-related medical or healthcare
nology leaders, and competitive landscape. Additionally, predicting the data; G16H20 with 713 patents includes ICT specially adapted for
innovation strengths of developing trends of all the patent applicants. therapies or health-improving plans, while G16H50 includes ICT
The patent applications are further mapped to the top CPC codes to specially adapted for medical diagnosis, medical simulation, or medical
review the technological analysis. data mining. The CPC group H04W with 2658 patents focuses on
Initial analysis presents, Samsung Electronics from South Korea wireless communication networks, H04W4 is bringing out new services
dominates the industry with 241 patents, and Korean companies leading specially adapted for wireless communication networks both from the
research in healthcare security technology is undisputed in contempo­ mobile application and wireless network domain.
rary times. Second, in lead is a major manufacturer of critical care de­
vices and systems, ZOLL medical corporation (acquired by Asahi Kasei 5. Methodology
group in April 2012) from the united states with 52 patents. Interna­
tional Business Machines Corporation, a software firm from the united The research builds on top of the recent innovations in machine
states follows with 24 patents. The fourth and fifth positions are held by learning (both supervised and unsupervised) for knowledge analytic
Capital One Service LLC and Medtronic Inc of the USA with 19 and 16 capabilities. This section outlines the comprehensive research frame­
patents. The remaining five include Elira, Inc., a group of three com­ work for intelligent collaborative patent mining, that combines sentient
panies namely Allied Security Trust, Siemens Healthcare GMBH, and computational intelligence. The primary input block is the Tech­
siemens medical solutions USA, Inc., another group of three companies nology domain architecture analysis, which reflects the current threat
Titan Health & Security Technologies, Inc.: WTI fund x, Inc.: venture state of a domain under investigation. The patents are reviewed using a
lending & leasing ix, Inc., Koninklijke Philips N.V., and Microsoft systematic framework consisting of two process parts where each part
emphasizes increasing the syntactic and semantic understanding of

6
U.H. Govindarajan et al. Computers & Security 133 (2023) 103404

Fig. 5. The evolution of the top 7 CPC (main) codes (left) along with the trend of patent applications vs publication (right).

syntactic and semantic understanding of telemedicine security patent

Table 3
space, and then mining specific threat domain knowledge using
Top 5 CPC (main) codes with their sub-technologies and patent count.
collaborative intelligence. The whole computational methodology is
Ranking CPC (main) CPC Patent Count Patent divided into 5 Blocks. The first block deals with pre-processing the text
(group) (CPC group) Count (CPC
and creating topics and probability scores from patent data. BERTopic is
main)
utilized in the generation of topics and probability scores. BERTopic is a
1 H04L: Transmission of H04L63 1112 3312
topic modeling approach that employs transformer architectural fea­
digital information H04L67 997
H04L5 690 tures and c-TF-IDF to generate high-density clusters that allow for
H04L9 567 explainable topics while retaining important words in topic de­
2 G16H: Healthcare G16H40 1366 2729 scriptions. The methodology is a combination of statistical modeling and
informatics G16H10 1250 deep learning with transformers.
G16H20 713
Fig. 7 shows the working of BERTopic. Bidirectional Encoder Rep­
G16H50 698
G16H80 476 resentations from Transformers (BERT) (Devlin et al., 2018) is used for
3 H04W: Wireless H04W4 933 2658 representing text data to high dimensional mathematical space. The
communication H04W72 822 Transformers model represented a significant advancement in NLP by
networks H04W76 601
significantly surpassing existing state-of-the-art frameworks across a
H04W12 466
4 G06F: Electric digital G06F21 1099 2381
swath of language modeling tasks. Transformers advance on recent work
data processing G06F16 489 on pre-training contextual representations, such as Semi-supervised
5 A61B: Diagnosis, A61B5 1042 1993 Sequence Learning (Dai and Le, 2015), Generative Pre-Training (Rad­
surgery, and A61B17 352 ford et al., 2018), ELMo (Peters et al., 2018), and ULMFit (Howard and
identification
Ruder, 2018). Transformers are the first deeply bidirectional, unsuper­
vised language representation, pre-trained using only a plain text
Telemedicine cyber security space. The first part is concerned with dataset (Wikipedia text corpora). The Transformers framework excels at
classifying the technology space into multiple threat layers from the capturing the fact that the meaning of a word can varies greatly
human intelligence standpoint. The second part is concerned with depending on the context, even within the same document or sentence.
computation and validation as presented in Fig. 6. The whole process (Devlin et al., 2018) Arora et al. has suggested the Transformers algo­
can be summarized mathematically as follow. Denote an input patent rithm is best suited for domains with huge volumes of training text and
dataset as Dinput. We assume there are N data items in Dinput, where language that is complicated with ambiguous uses that can be very
Dinput = (d1, …, dN). The methodology is divided into five process context-specific (Arora et al., 2020). Patents are an appropriate domain
blocks. (a) The input patent data is passed through BERTopic (Groo­ for the use of the Transformers algorithm. Furthermore, synonyms are
tendorst, 2022) to generate topic ti ∈ T and probability scores pi ∈ P of an important aspect of patent operations because they form the foun­
those topics. (b) Each patent data points di is assigned to one of the dation of prior art searching. Srebovic et al. further suggests that the
topics T with the highest topic probability. (c) Topics are segmented into BERT method should be particularly useful because it is sufficiently
identified threats in the previous section, resulting in patent data cluster flexible in generating a list of synonyms for any given term while ac­
ci ∈ C for each threat subspace. (d) Extrinsic and intrinsic validation of counting for its surrounding context (Srebrovic, 2019). Uniform Mani­
each cluster ci ∈ C. (e) An analysis is carried out on each ci to understand fold Approximation and Projection (UMAP) (McInnes et al., 2018) is
the solution space for the ith threat. used for dimensionality reduction and Hierarchical Density-based
The process comprises broadcasting the patent text space to a high spatial clustering of applications with noise (HDBSCAN) (Malzer and
dimensional vector and thus generating the topic space to capture the Baum, 2020) is used for clustering to generate similar clusters. Finally,

7
U.H. Govindarajan et al. Computers & Security 133 (2023) 103404

Fig. 6. Computation Flow Diagram.

Fig. 7. BERTopic Algorithm Flow.

class-based TF-IDF (c-TF-IDF) (Xu and Wu, 2014) is used to generate Incopat global patent database investigating the research direction to­
candidates by extracting class-specific words. ward Cyber Security Considerations for Telehealth. The proposed
In the second block, each patent data is assigned to the topic with the methodology allows the grouping and analysis of similar patents from a
highest probability. This way we create a patent data cluster for each large dataset enhancing the traditional manual methods of exploration.
topic space. The third block deals with the segmentation of topic space The validation of the methodology is accomplished in two steps: the first
into 7 identified threat spaces. The fourth block deals with cluster validates the Topic model, and the second validates the topic grouping
validation. Finally, the fifth block constitutes the analysis of each cluster within threats. The Topic Coherence C_v metric is used to validate the
space. topic modeling. Topic Coherence scores a single topic by calculating the
degree of semantic similarity between the topic’s high-scoring words.
6. Validations and findings These measurements aid in distinguishing between topics that are
semantically interpretable and topics that are statistical inference arti­
The section logically extends to the analysis of patent data from the facts. The C_v metric is based on a sliding window, one-set segmentation

8
U.H. Govindarajan et al. Computers & Security 133 (2023) 103404

of the top words, and an indirect confirmation measure that employs device that provides interaction-based knowledge sharing of each stu­
normalized pointwise mutual information (NPMI) and cosine similarity. dent and the professor, an appraisal history/guidance history of the
The distribution of coherence score concerning min cluster size in professor in terms of immediate feedback (emergency measures, pre­
HDBScan is analyzed to determine the number of topics to be explored. scription) of the professor immediately transmitted to the user terminal,
Table 4 shows the distribution of topic coherence concerning the min a training history of the learner/trainee, UBT/SBT test and marking,
cluster size in HDBScan. appraisal of the processor and administrative control and, also, sys­
The proposed methodology identified 45 topics with a coherence temically manages the same. The awareness of Health Information Ex­
score of 0.610 with a minimum topic size of 34 documents per topic. change (HIE) security measures and sharing procedures encourages
Fig. 8 shows one topic from each threat space and its word probability patients to be cognitively and emotionally involved with the HIE system
distribution. (Esmaeilzadeh, 2020).
The topic grouping in various threats is validated by a thorough The cluster telemedicine devices consist of patent grants with claims
analysis of patents within each threat space discussed below in this on phones, home computers, and IoT devices that have access to medical
section. The 45 topics obtained from the topic modeling process were PHI information. The patent in this cluster mainly deals with secure
distributed amongst the defined threat points. out of these 45 topics 28 mobile devices framework with data consistency and IoT devices data
significant topics were identified and assigned amongst the 7 threat exchange. Data from digital health devices requires standard commu­
spaces. Threat 8 is named for a topic with the highest frequency of patent nication protocols so that it can be used by any authorized data collec­
documents and looked at securing more than one failure point in the tor. Data should not be restricted to a single vendor. Devices must have
telemedicine infrastructure. Table 5 highlights the distribution of clear criteria for evaluating data from these devices, and regulatory
several topics and patent data within each threat and the words from organizations should develop criteria for these devices’ data (Condry
topics within each threat space. and Quan, 2021). Samsung Electronics Co Ltd has several patents in this
The patent frequency analysis in Table 6 demonstrates that perhaps cluster. US9973505B2 is a patent granted to Samsung Electronics Co Ltd
the threat of patient vulnerability (1), which collectively accounts for which discloses A method for controlling contents and an electronic
the highest number of security infringements, seems to have the fewest device. The method provides an operation method for setting access to
patent data, accompanied by Telemedicine service provider threat (7), contents and managing that access mechanism for securing the content
which is also an evolving trend that requires substantial security access. US20160330573A1 is a patent granted to Sorin CRM SAS that
consideration. Threat 6 (telemedicine service provider) and Threat 8 provides a medical monitoring and communication system for wireless
(the entire telemedicine infrastructure) account for the largest quantity communication between an implantable medical device, and a mobile
of patents. Table 7 below displays the top CPCs and companies that have user device. Another invention application CN111431922A from China
the most publications for each threat space. Construction Bank Corp and CCB Fintech Co Ltd provides an Internet of
Creating a smart healthcare organization through the use of both Things data encryption transmission method and system for securing
offline and online technological solutions, as well as bolstering part­ sensitive data in IoT devices such as wearables. Patent application
nerships and collaborations from multiple sectors of the health industry JP2016201150A by Synchronous Technologies INC discusses systems
(e.g., healthcare devices, pharmaceutical, insurance, system develop­ and methods for a secure mobile framework to securely connect appli­
ment, and personal information security companies) (Lee and Lee, cations running on mobile devices to services within an enterprise.
2021). The research further proposes standardization of these partner­ There was very little activity in the area of spyware and malware
ships into the security infrastructure. All the threats are analyzed further analysis and detection.
for a deep dive patent point of view. The other emerging areas of interest The cluster home network consist of patent grants with claims on
are holistic infrastructure security including physical security and a securing and evaluating network infrastructure. Evaluation of the home
strong indication of blockchain usage for health data sharing that form network is a very necessary step to secure it. Autonomous diagnostic of
the basis of further investigation. These threat spaces are a major area of computer networks is a major domain in this cluster.
exploration and are corroborated by interaction with industry pro­ US20200358800A1 invention application from Akitra Inc discloses an
fessionals who are focusing on them a lot as they have a direct impact on artificial intelligence-based system and method to detect anomalies in
health data security in telemedicine the computer network. The system comprises an enterprise network
Patient vulnerability is one of the key factors leading to the rise in system and engine. The engine has an AI and machine learning-based
data breaches (Warburton, 2020). Virtual health consulting and remote monitoring and detection module and a remediation module coupled
health management apps may be rendered ineffective in the presence of to the switch device. The remediation module is configured to initiate a
citizens with limited digital literacy. Bridging the digital health literacy remediation process based upon the detection of at least one of the
gap is especially important because digital knowledge gaps typically anomalies from the flow of data. Another patent CN106850551A from
affect those who are most in need of health care (for example, the Changchun Science Tech university discloses a network security risk
elderly) (Raimo et al., 2022). The only way to solve this vulnerability is evaluation and autonomic defense system. The methodology provides
to educate the population about the infrastructure and technology. The real-time monitoring and auditing of the network traffic and the ability
main challenge in this solution is information overload and information to restore the network autonomously.
reach (Govindarajan et al., 2022). Major patents granted in this cluster The Gateway cluster consists of 132 patent grants with claims on
are for increasing the reach of education efficiently to facilitate educa­ secure communication with medical devices, systems for managing,
tion at all levels without affecting the learning process. An example case controlling, and monitoring medical devices, and secure healthcare
is KR Patent grant KR1020200007984A granted to Cetalab co Ltd which gateways. The patent JP6884753B2 granted to Inspire Medical Systems
discloses a method of ubiquitous-based training education portfolio Inc proposes a communication platform that establishes secure com­
system. The ubiquitous-based training education portfolio system pro­ munications between a medical device and a trusted authority (TA)
vides a cloud training education management service (U-portfolio) using service provider. Secure communications prevent the communication
big data analysis and artificial intelligence techniques using a smart platform from accessing the data while allowing access to the data at the

Table 4
Topic coherence score C_v concerning different minimum cluster size.
Minimum Cluster size 10 14 18 22 26 30 34 38 42 46 50

Coherence Score 0.497 0.491 0.467 0.519 0.547 0.522 0.610 0.547 0.539 0.469 0.534

9
U.H. Govindarajan et al. Computers & Security 133 (2023) 103404

Fig. 8. Topic keyword probability distribution across threat space.

medical device and/or the trusted authority service provider. Another granted CN107733565B patent to Samsung Electronics Co Ltd which
patent US20180013832A1 from Electronics and Telecommunications discloses a method for decoding by a terminal in a mobile communi­
Research Institute discloses a method for securing a protocol using the cation system with security considerations. It is done by receiving
health device and the gateway device. Moreover, the authentication and setting information on whether a second channel code which is an outer
key exchange is based on security session information. code for a first channel code is set or not from the base station and
The Internet (public network) consists of 394 patent grants with performing the decoding based on a result of the determined. Another
claims on encoding and decoding of data in communication systems, patent from Samsung Electronics Co Ltd KR1020200033081A, focuses
secure transmission, and reception of signals, and methods of security on a method for allowing a base station to check an encryption key
key generation. Samsung Electronics Co Ltd has several patents in this (COUNT value) for enhanced security.
cluster. China National Intellectual Property Administration (CNIPA) The Telemedicine system consists of 966 patent grants with claims

10
U.H. Govindarajan et al. Computers & Security 133 (2023) 103404

Table 5
Topic count and patent count and words from topics in each threat.
Threat 1 2 3 4 5 6 7 8
Topic count 1 7 3 2 4 10 1 1
Patent count / 100 0.38 4.46 2.63 1.35 3.94 9.76 0.69 56.36
Threats Words from grouped topics
Threat #1: Patients Vulnerability student, training, campus, management, education, professor, school, competition, portfolio, learnertrainee
Threat #2: Telemedicine devices peripheral, machine, device, mode, content, control, electronic, slave unit, code, present disclosure, device, external, medical,
ambulatory medical, configure, implantable, signal, first, implant sensor, measurement, monitoring, monitor, electrode, measure,
patient, wearable, user authentication application access secure datum identity key method data communication, technology, mobile,
convergence, efficiently transmitted, hospital interrogator, system, transmission rate, scheme, supporting higher, city smart, device,
communication, interface, second, processor, screen, battery, least, host, server, implement, healthcare, deidentified, carrier container,
data there, anonymous
Threat #3: Home network intelligent service, technology, communication, present disclosure, iotrelate, safety related, connected car, home smart, car,
communication technique, network, security, risk, vulnerability, entity, score, operational, threat, datum, analysis, pdcp, technology,
basis, terminal, area, system, technology, smart, generation, smart build, smart home, retail, generation
Threat #4: Gateway devices communication, medical, gateway, device, secure, intravascular, wave, processor, access, key
Threat #5: Internet (public network) base station, communication, transmit, technology, basis, terminal, signal, receive, transmission, step, lightconnection, radio, ue, access,
presentdisclosure, network, connection, smart, control, basestation, processing, signaltransmitte, signalgenerate, thereof, present,
copyrightkipo, supporthigher, fuse, servicesetc, transmissionrate, citysmart, buildingsmart, homesmart
Threat #6: Telemedicine system customer, identify, management, information, service, provide, condition, device, network, technology, functionalclassification,
intelligent, community, maintenance, power, module, operation, platform, system, city, blockchain, medical, sharing, datum,privacy,
record,storage,base,share,contract, medical, hospital, management, health, information, patient, insurance, payment, socialsecurity,
card, settlement, doctorseee, accord, credit, application, cloud, deployment, software, cluster, resource, prescription, drug, medicine,
medication, pharmacy, electronic, dispense, pharmacist, cloud, ciphertext, encryption, search, encrypt, access, privatekey, information,
identification, acquire, display, history, desk, menu, terminal, card, cardreade, cardreader, box, self, equipment, screen
Threat #7: Telemedicine service provider virtualreality, video, image, realitystreame, adaptivevirtual, scene, service, display, screen, pointer
Threat #8: Two or more threats resolved security, system, information, datum, device, method, communication, medical, patient, user, provide, B8,

on PHI data storage and security using blockchain, telemedicine plat­ aggregated data from multiple static sources, data that is immune to
forms, secure healthcare infrastructure, and a secure dispersed network tampering and protection, and lower cyber risk (Cerchione et al., 2023).
for improved communications between healthcare industry partici­ The cluster “Two or more threats resolved” with the largest number
pants. The patent KR101962541B1 discloses a device and a method for of patents was initially thought of as a noise cluster. Further investiga­
controlling medical information and its access. Another patent tion revealed that the patent within this cluster focuses on securing total
IN202041008020A focuses on the design of healthcare network archi­ processes within telemedicine infrastructures like mobile telemedicine
tecture for the dissemination of e-health records using blockchain units, methods for enforcing endpoint health security standards, and
technology. CN111400401A from Jiangsu Rongzer Inf Tech Co Ltd cryptography. The patent US10181011B2 granted to Epic Systems
provides an electronic medical record storage system based on block­ Corporation provides a method for a healthcare information system to
chain comprising login management, medical record management, and securely exchange clinical information by using at least one clinical
a blockchain client for access. The patent US20210174972A1 from exchange server on the network. Another patent US11005661B1 gran­
PatientMD Inc discusses a secure dispersed network for improved ted to KPN Innovations LLC proposes a system for cryptographically
communications between healthcare industry participants. Researchers secured outputs from telemedicine sessions.
have expressed a strong interest in the use of blockchain in healthcare,
whether for health information exchange, healthcare provider networks, 7. Conclusions implications and future scope
patient-generated health data, or a variety of other applications, how­
ever, remain several research and operational challenges, when This article identifies several promising topics that are relevant to the
attempting to fully integrate blockchain technology with existing EHR advancement of Telemedicine growth but are also applicable to adjacent
systems (Shi et al., 2020). The rise in blockchain technology adaptation areas of study, such as network security, secure data sharing, and health
research can be seen not only in academic research highlighting the data privacy, in which innovation is a major driver of survival, renewal,
existing and future landscape of the technology but also in industries and success. Accelerated telemedicine growth will benefit not only
and governments appreciating the potential of this technology as a developed countries but will also have a significant social impact on
secure and reliable means of operation in the virtual and physical worlds treatment in developing countries (Biancone et al., 2021). Given that
(Daim et al., 2020). our study differs fundamentally from previous literature reviews that
The Telemedicine service provider consists of 68 patent grants with examine Telemedicine security patents more broadly, it provides guided
claims on secure messaging and facilitating video/audio/VR-based division and visualization of patent space into the threats space identi­
interaction between two systems. The patent US10397297B2 granted fied within the entire telemedicine infrastructure, empowering practi­
to Samsung Electronics Co Ltd discloses a method for receiving image tioners with numerous recommendations for adopting and governing
data for a virtual-reality streaming service by an electronic device based innovation-based Telemedicine infrastructure. Jalali et al. (2021)
on bandwidth state and head-tracking information and bandwidth in­ concluded in their research that during the shift from traditional ways to
formation collected in advance. Another patent CN107368697B granted telemedicine, new issues and risks will unravel that need to be
to Shanghai United Imaging Healthcare Ltd discloses a virtual reality addressed, particularly regarding information security and privacy, and
control method of medical equipment, which might be especially useful ongoing work is needed to ensure that our technology infrastructure
in a pandemic-like situation. The patent IN202041021694A proposes a provides an environment for safe and effective care delivery. This
secure direct messaging and health information exchange (hie) using comprehensive study focuses on reviewing and synthesizing the frag­
blockchain technology. The method provides high security against data mented intersection of telehealth security requirements and innovation.
breaches and attacks and provides trust between the patients and It outlines theoretical underpinnings, major patent themes, accepted
various healthcare professionals over existing methodologies. The use of methodologies, and levels of analysis. And demonstrated limited
blockchain will be critical in the design of healthcare systems. It could knowledge of existing solutions in the field of telehealth security. Kau­
lead to automated data collection and authentication, accurate and tish et al. (2023) point out trust as an important parameter which

11
U.H. Govindarajan et al. Computers & Security 133 (2023) 103404

Table 6 The focus on educating the user which is of extreme importance needs
Topic count and patent count and words from topics in each threat. more attention. Blockchain technology emerged as one of the most
Top Assignee Threat Top prominent solutions for medical data transactions, storage, and sharing.
CPC’s However, beyond privacy and security, there are various significant
Say Yes to Education, INC. Threat #1: Patients G06Q50 blockchain challenges that we need to confront as an industry, such as
Vulnerability G06Q10 the environmental and financial costs associated with implementing
G07C9 these systems (Massaro, 2021). These aspects require further research
G09B5 however, the proposed framework due to its generic nature allows for its
Zoll Medical Corporation Threat #2: Telemedicine A61B5
Samsung Electronics CO., LTD. devices H04L63
application in other domain niche explorations in telemedicine and
International Business Machines H04L9 eHealth. The methodology serves as an in-depth template that helps
Corporation researchers and practitioners gain a better understanding of existing
Masimo Corporation G06F21 technologies and their potential applications. It serves as a valuable
G16H40
resource for developing new security measures and helping interest
Samsung Electronics CO., LTD. Threat #3: Home network H04W72
Albeado, INC. H04L5 groups like telehealth providers, healthcare professionals and patients.
Level 3 Communications, LLC H04L63 We outline theoretical underpinnings, major patent themes, accepted
Microsoft Technology Licensing, LLC H04L1 methodologies, and levels of analysis, and primarily explored techno­
H04W76 logical direction in the telemedicine security space based on a system­
Samsung Electronics CO., LTD. Threat #4: Gateway H04W12
devices A61B5
atic review of a large corpus of global patent grants from over 40 key
Zoll Medical Corporation H04L63 patent office’s implying an intelligent collaborative patent mining
H04L9 which represents intelligence from both the human and computational
Hill-rom Services, INC. A61N1 domains. The patent space is divided into seven major threats using a
H04W8
topic generation methodology based on transformer architecture and
Samsung Electronics CO., LTD. Threat #5: Internet (public H04W72
network) H04L5 LDA, allowing for a more accurate assessment of the threat and solution
H04W76 space as well as further refining the technology domain and its emerging
H04L1 topics. This comprehensive overview opens new avenues of research for
Research & Business Foundation H04W36 scholars interested in telemedicine security. Further, policymakers and
Sungkyunkwan University H04W74
regulatory bodies can utilize the findings to inform the development of
H04W52
H04W4 telehealth regulations and guidelines. The major factor of concern for
Electronic Health Record Data, INC. Threat #6: Telemedicine H04L67 the patient i.e., data privacy can only be addressed through measures
Synerio Technologies, INC. system G06F21 such as encryption, secure data transmission, and access controls
AT&T Intellectual Property I, L.P. G06Q10
implemented to safeguard data confidentiality (Vesselkov et al., 2018).
General Electric Company G16H10
Samsung Electronics CO., LTD. G16H40 Therefore, this study serves as a template to provide an up-to-date check
The Quantum Group, INC. H04L63 on emerging updates to security measures that are essential to address
Samsung Electronics CO., LTD. Threat #7: Telemedicine H04N21 evolving threats. The accelerated growth of telemedicine will not only
Skreens Entertainment Technologies, service provider G06F3 benefit developed countries but also have a significant social impact on
INC.
treatment in developing countries. By providing a guided division and
Siemens Healthcare Diagnostics INC. H04L65
Scenera, INC. H04L67 visualization of the patent space into identified threats within the entire
State Farm Mutual Automobile H04N7 telemedicine infrastructure, this study empowers practitioners with
Insurance Company numerous recommendations for adopting and governing
Triad national security, LLC G06T3
innovation-based telemedicine infrastructure. Despite its contributions,
Samsung Electronics CO., LTD. Threat #8: Two or more A61B5
Siemens Healthcare GMBH threats resolved H04L63
the review has some shortcomings. The methodological approach,
Capital One Services, LLC H04L67 although robust, is subject to author bias due to human intelligence
Zoll Medical Corporation G06F21 input. Furthermore, the holistic approach, while valuable and compre­
International Business Machines H04L5 hensive, does not allow for an in-depth analysis of the identified patents
Corporation
in each threat. Nonetheless, the study aims to influence and spark
Medtronic, INC. H04W4
Ice Cap, series 106 of Allied Security G16H40 additional research on the topic of telemedicine security and adoption
Trust I roadmap, providing a novel and clear roadmap for further exploration.
Owl Rock Capital Corporation, as H04W72
Collateral Agent
Declaration of Competing Interest
prevents resistance to adoption and facilitates the entry of newly sug­
The authors declare that they have no known competing financial
gested telemedicine products into the market. Our investigation is
interests or personal relationships that could have appeared to influence
helpful to the stakeholders involved in the space giving them crucial
the work reported in this paper.
considerations on data privacy and telemedicine security
considerations.
Data availability
Furthermore, the methodology employs a novel topic generation
technique that emphasizes both topic and contextual representation of
Data will be made available on request.
natural language, which has not previously been used in the patent
literature mining space. Although the proposed fully adaptive value
evaluation process is limited to Telemedicine security patent valuation
analysis and verification, the methodology is generic and can be applied Acknowledgements
in other domain sectors. This analysis reveals that telehealth and secu­
rity is a recognized problems with focused participation by a variety of This research was supported by the National Natural Science Foun­
industries. Except for one, every threat space has numerous patents that dation of China [Grant 52122215], Shanghai Shuguang Program [Grant
discuss the solution space to make telemedicine security more robust. 22SG45], and the University of Shanghai for Science and Technology,
Business School Young Teacher Research Start-up Fund.

12
U.H. Govindarajan et al. Computers & Security 133 (2023) 103404

References Monaghesh, E., Hajizadeh, A., 2020. The role of telehealth during COVID-19 outbreak: a
systematic review based on current evidence. BMC Public Health 20 (1), 1–9.
Peters, M. E., Neumann, M. E., Iyyer, M., Gardner, M., Clark, C. M., Lee, K., &
Anwar, Z., Bibi, N., Afzal, H., 2019. Mining mobile security concerns and their solutions.
Zettlemoyer, L. (2018). Deep contextualized word representations. arXiv (Cornell
In: Proceedings of the 2019 16th International Bhurban Conference on Applied
University). arXiv:1802.05365.
Sciences and Technology (IBCAST), pp. 640–646.
Radford, A., Narasimhan, K., Salimans, T., & Sutskever, I. (2018). Improving language
Arora, S., May, A., Zhang, J., & Ré, C. (2020). Contextual embeddings: when are they
understanding by generative pre-training.
worth it?. arXiv preprint arXiv:2005.09117.
Raimo, N., De Turi, I., Albergo, F., Vitolla, F., 2022. The drivers of the digital
Baltzan, P., Phillips, A., 2014. Ebook: Business Driven Information Systems. McGraw
transformation in the healthcare industry: an empirical analysis in Italian hospitals.
Hill.
Technovation, 102558.
Biancone, P., Secinaro, S., Marseglia, R., Calandra, D., 2021. E-health for the future.
Rezaei, M., Jafari-Sadeghi, V., Cao, D., Mahdiraji, H.A., 2021. Key indicators of ethical
Managerial perspectives using a multiple case study approach. Technovation,
challenges in digital healthcare: a combined Delphi exploration and confirmative
102406.
factor analysis approach with evidence from Khorasan province in Iran. Technol.
Bokolo, A.J., 2021. Exploring the adoption of telemedicine and virtual software for care
Forecast. Soc. Change 167, 120724.
of outpatients during and after COVID-19 pandemic. Irish J. Med. Sci. (1971) 190
Ribeiro-Navarrete, S., Saura, J.R., Palacios-Marqués, D., 2021. Towards a new era of
(1), 1–10.
mass data collection: assessing pandemic surveillance technologies to preserve user
Buvik, A., Bugge, E., Knutsen, G., Småbrekke, A., Wilsgaard, T., 2016. Quality of care for
privacy. Technol. Forecast. Soc. Change 167, 120681.
remote orthopaedic consultations using telemedicine: a randomised controlled trial.
Salahdine, F., Kaabouch, N., 2019. Social Engineering Attacks: a Survey. Future Internet
BMC Health Serv. Res. 6 (1), 1–11.
11, 89.
Cegarra-Sánchez, J., Cegarra-Navarro, J.G., Chinnaswamy, A.K., Wensley, A., 2020.
Seh, A.H., Zarour, M., Alenezi, M., Sarkar, A.K., Agrawal, A., Kumar, R., Ahmad Khan, R.,
Exploitation and exploration of knowledge: an ambidextrous context for the
2020. Healthcare data breaches: insights and implications. In: Healthcare, 8.
successful adoption of telemedicine technologies. Technol. Forecast. Soc. Change
Multidisciplinary Digital Publishing Institute, p. 133.
157, 120089.
Sheingold, B.H., Hahn, J.A., 2014. The history of healthcare quality: the first 100 years
Cerchione, R., Centobelli, P., Riccio, E., Abbate, S., Oropallo, E., 2023. Blockchain’s
1860–1960. Int. J. Afr. Nurs. Sci. 1, 18–22.
coming to hospital to digitalize healthcare services: designing a distributed
Shi, S., He, D., Li, L., Kumar, N., Khan, M.K., Choo, K.K.R., 2020. Applications of
electronic health record ecosystem. Technovation, 120,, 102480.
blockchain in ensuring the security and privacy of electronic health record systems: a
Condry, M.W., Quan, X.I., 2021. Digital health innovation, informatics opportunity, and
survey. Comput. Secur. 97, 101966.
challenges. IEEE Eng. Manage. Rev. 49 (2), 81–88.
Shokri, T., Lighthall, J.G., 2020. Telemedicine in the era of the COVID-19 pandemic:
Dai, A.M., Le, Q.V., 2015. Semi-supervised sequence learning. Adv. Neural Inf. Process
implications in facial plastic surgery. Facial Plast. Surg. Aesthet. Med. 22 (3),
Syst. 28.
155–156.
Daim, T., Lai, K.K., Yalcin, H., Alsoubie, F., Kumar, V., 2020. Forecasting technological
Soltanisehat, L., Alizadeh, R., Hao, H., Choo, K.K.R., 2020. Technical, temporal, and
positioning through technology knowledge redundancy: patent citation analysis of
spatial research challenges and opportunities in blockchain-based healthcare: a
IoT, cybersecurity, and Blockchain. Technol. Forecast. Soc. Change 161, 120329.
systematic literature review. IEEE Trans. Eng. Manage.
Dash, S., Shakyawar, S.K., Sharma, M., Kaushik, S., 2019. Big data in healthcare:
Sönmez, F.Ö., Hankin, C., Malacaria, P., 2022. Decision support for healthcare cyber
management, analysis and future prospects. J. Big Data 6 (1), 1–25.
security. Comput. Secur. 122, 102865.
Devlin, J., Chang, M.W., Lee, K., & Toutanova, K. (2018). Bert: pre-training of deep
Srebrovic, R. (2019). Expanding your patent set with ML and BigQuery.
bidirectional transformers for language understanding. Preprint arXiv:1810.04805.
Trappey, A.J., Trappey, C.V., Govindarajan, U.H., Sun, J.J., 2019. Patent value analysis
Esmaeilzadeh, P., 2020. The effect of the privacy policy of Health Information Exchange
using deep learning models—The case of IoT technology mining for the
(HIE) on patients’ information disclosure intention. Comput. Secur. 95, 101819.
manufacturing industry. IEEE Trans. Eng. Manage. 68 (5), 1334–1346.
Galetsi, P., Katsaliaki, K., Kumar, S., 2022. Exploring benefits and ethical challenges in
Vesselkov, A., Hämmäinen, H., Töyli, J., 2018. Technology and value network evolution
the rise of mHealth (mobile healthcare) technology for the common good: an
in telehealth. Technol. Forecast. Soc. Change 134, 207–222.
analysis of mobile applications for health specialists. Technovation, 102598.
Warburton, D. (2020). Phishing Attacks Soar 220% During COVID-19 Peak as
Garcia-Perez, A., Cegarra-Navarro, J.G., Sallos, M.P., Martinez-Caro, E.,
Cybercriminal Opportunism Intensifies. Www.f5.com. https://www.f5.com/compa
Chinnaswamy, A., 2023. Resilience in healthcare systems: cyber security and digital
ny/news/features/phishing-attacks-soar-220–during-covid-19-peak-as-cybercrimina
transformation. Technovation 121, 102583.
l.
Gaurav, A., Gupta, B.B., Panigrahi, P.K., 2022. A novel approach for DDoS attacks
Wosik, J., Fudim, M., Cameron, B., Gellad, Z.F., Cho, A., Phinney, D., Tcheng, J., 2020.
detection in COVID-19 scenario for small entrepreneurs. Technol. Forecast. Soc.
Telehealth transformation: COVID-19 and the rise of virtual care. J. Am. Med.
Change 177, 121554.
Inform. Assoc. 27 (6), 957–962.
Geum, Y., Kim, M., 2020. How to identify promising chances for technological
Xu, D.D., Wu, S.B., 2014. An improved TFIDF algorithm in text classification. In: Applied
innovation: keygraph-based patent analysis. Adv. Eng. Inf. 46, 101155.
Mechanics and Materials, 651. Trans Tech Publications Ltd, pp. 2258–2261.
Govindarajan, U.H., Narang, G., Kumar, M., 2022. Graphic facilitation in the engineering
Yaqoob, I., Ahmed, E., ur Rehman, M.H., Ahmed, A.I.A., Al-garadi, M.A., Imran, M.,
workflow: adoption framework, barriers, and future roadmap. IEEE Eng. Manage.
Guizani, M., 2017. The rise of ransomware and emerging security challenges in the
Rev.
Internet of Things. Comput. Netw. 129, 444–458.
Grootendorst, M. (2022). BERTopic: neural topic modeling with a class-based TF-IDF
Yuan, N., Dudley, R.A., Boscardin, W.J., Lin, G.A., 2019. Electronic health records
procedure. Preprint arXiv:2203.05794.
systems and hospital clinical performance: a study of nationwide hospital data.
Hassan, S.S., 2018. Soumik Das Bibon, Md Shohrab Hossain, and Mohammed
J. Am. Med. Inform. Assoc. 26 (10), 999–1009.
Atiquzzaman. Security threats in bluetooth technology. Comput. Secur. 74
(308–322), 72.
Howard, J., & Ruder, S. (2018). Universal language model fine-tuning for text
classification. Preprint arXiv:1801.06146. Dr. Usharani Hareesh Govindarajan is an Assistant Professor
Jalali, M.S., Landman, A., Gordon, W.J., 2021. Telemedicine, privacy, and information of Information Management and Systems at the Business
security in the age of COVID-19. J. Am. Med. Inform. Assoc. 28 (3), 671–672. School, University of Shanghai for Science and Technology,
Kautish, P., Siddiqui, M., Siddiqui, A., Sharma, V., Alshibani, S.M., 2023. Technology- Shanghai, China. He received his Doctoral degree in Industrial
enabled cure and care: an application of innovation resistance theory to telemedicine Engineering from National Tsing Hua University, a Master’s in
apps in an emerging market context. Technol. Forecast. Soc. Change 192, 122558. Engineering in Computer Science, and a Master’s in Science in
Khansa, L., Cook, D.F., James, T., Bruyaka, O., 2012. Impact of HIPAA provisions on the Informatics from the University of Delhi, India. Hareesh’s
stock market value of healthcare institutions, and information security and other research interests include technology management for inter­
information technology firms. Comput. Secur. 31 (6), 750–770. national business with a focus on analytics design, low-code
Kichloo, A., Albosta, M., Dettloff, K., Wani, F., El-Amir, Z., Singh, J., Chugh, S., 2020. tools, and open data-related frameworks in enabling decision
Telemedicine, the current COVID-19 pandemic and the future: a narrative review support. Further, he is a member of several industrial-academic
and perspectives moving forward in the USA. Family Med. Community Health 8 (3). initiatives across China, India, and France and has authored
Lee, S.M., Lee, D., 2021. Opportunities and challenges for contactless healthcare services key research papers in leading international journals such as
in the post-COVID-19 Era. Technol. Forecast. Soc. Change 167, 120712. IEEE Transactions on Engineering Management, Complexity, IEEE Engineering Manage­
Luciano, E., Mahmood, M.A., Mansouri Rad, P., 2020. Telemedicine adoption issues in ment Review, Advanced Engineering Informatics, etc. Readers can follow updates and
the United States and Brazil: perception of healthcare professionals. Health engage further on LinkedIn @Hareeshpillai.
Informatics J. 26 (4), 2344–2361.
Malzer, C., Baum, M., 2020. A hybrid approach to hierarchical density-based cluster
selection. In: Proceedings of the 2020 IEEE International Conference on Multisensor
Fusion and Integration for Intelligent Systems (MFI). IEEE, pp. 223–228.
Mamdouh, M., Awad, A.I., Khalaf, A.A., Hamed, H.F., 2021. Authentication and identity
management of IoHT devices: achievements, challenges, and future directions.
Comput. Secur. 111, 102491.
Massaro, M., 2021. Digital transformation in the healthcare sector through blockchain
technology. Insights from academic research and business developments.
Technovation, 102386.
McInnes, L., Healy, J., & Melville, J. (2018). Umap: uniform manifold approximation and
projection for dimension reduction. Preprint arXiv:1802.03426.

13
U.H. Govindarajan et al. Computers & Security 133 (2023) 103404

Mr. Dhiraj Kumar Singh is a Data Scientist at Edifecs, USA. He Dr. Hardik Gohel is an Assistant Professor & Director of
earned his master’s degree in Informatics in 2019 from the Research, in the Department of Computer Science, Director of
Institute of Informatics and Communication, University of the Applied AI Research Laboratory at the University of
Delhi, India. He has been working in the data science industry Houston–Victoria, USA. Dr. Gohel has extensive research
as a data scientist for the past three years. He is currently experience in artificial intelligence and his research projects
serving Edifecs, USA, where his research focuses on the have involved cyber test automation and monitoring, smart
extraction and analysis of patient risk from historical medical bandages for wound monitoring, big data for security intelli­
data for quick interventions. His research interest also includes gence, trustworthy cyberspace for security and privacy of so­
the use of neural networks and explainable AI in health care cial media, predictive maintenance for nuclear infrastructure,
systems. and database and mobile forensics infrastructure.

14