JMIR AI Benjamens et al

Original Paper

Forecasting Artificial Intelligence Trends in Health Care:

Systematic International Patent Analysis

Stan Benjamens1, MD, PhD; Pranavsingh Dhunnoo2,3, MD, MSc; Márton Görög2, MSc; Bertalan Mesko2, MD, PhD
1
Department of Surgery, Ikazia Hospital, Rotterdam, Netherlands
2
The Medical Futurist Institute, Budapest, Hungary
3
Department of Computing, Donegal Campus, Atlantic Technological University, Letterkenny, Ireland

Corresponding Author:
Bertalan Mesko, MD, PhD
The Medical Futurist Institute
Povl Bang-Jensen u 2/B1 4/1
Budapest, 1118
Hungary
Phone: 36 703807260
Email: [email protected]

Abstract
Background: Artificial intelligence (AI)– and machine learning (ML)–based medical devices and algorithms are rapidly
changing the medical field. To provide an insight into the trends in AI and ML in health care, we conducted an international
patent analysis.
Objective: It is pivotal to obtain a clear overview on upcoming AI and MLtrends in health care to provide regulators with a
better position to foresee what technologies they will have to create regulations for, which are not yet available on the market.
Therefore, in this study, we provide insights and forecasts into the trends in AI and ML in health care by conducting an international
patent analysis.
Methods: A systematic patent analysis, focusing on AI- and ML-based patents in health care, was performed using the Espacenet
database (from January 2012 until July 2022). This database includes patents from the China National Intellectual Property
Administration, European Patent Office, Japan Patent Office, Korean Intellectual Property Office, and the United States Patent
and Trademark Office.
Results: We identified 10,967 patents: 7332 (66.9%) from the China National Intellectual Property Administration, 191 (1.7%)
from the European Patent Office, 163 (1.5%) from the Japan Patent Office, 513 (4.7%) from the Korean Intellectual Property
Office, and 2768 (25.2%) from the United States Patent and Trademark Office. The number of published patents showed a yearly
doubling from 2015 until 2021. Five international companies that had the greatest impact on this increase were Ping An Medical
and Healthcare Management Co Ltd with 568 (5.2%) patents, Siemens Healthineers with 273 (2.5%) patents, IBM Corp with
226 (2.1%) patents, Philips Healthcare with 150 (1.4%) patents, and Shanghai United Imaging Healthcare Co Ltd with 144 (1.3%)
patents.
Conclusions: This international patent analysis showed a linear increase in patents published by the 5 largest patent offices. An
open access database with interactive search options was launched for AI- and ML-based patents in health care.

(JMIR AI 2023;2:e47283) doi: 10.2196/47283

KEYWORDS
artificial intelligence; patent; healthcare; health care; medical; forecasting; future; AI; machine learning; medical device;
open-access; AI technology

clinical decision-making to diagnostic imaging interpretation

Introduction [1,2]. Both the commercial development and academic research
Artificial intelligence (AI), in the form of machine learning focusing on AI and ML in health care showed an exponential
(ML)–based medical devices and algorithms, has been rapidly growth; however, regulation for clinical use and commercial
changing a range of aspects of the medical profession from rollout follow a slower path.

https://ai.jmir.org/2023/1/e47283 JMIR AI 2023 | vol. 2 | e47283 | p. 1

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR AI Benjamens et al

The US Food and Drug Administration (FDA) has been leading

the way for regulators worldwide, being the first regulatory
Methods
body to adopt an AI policy and provide guidelines for approving Selection of Patents
AI-based medical technologies in practice [3].
We selected the Espacenet search engine of the EPO to access
Certain medical specialties stand out in terms of the impact of the data from the 5 international patent offices collaborating in
AI on the practice of those professions. Based on a previous the Global Dossier initiative [8-10]. The Global Dossier
study by our research group, examples include cardiology, initiative enables web-based public access for the patent data
radiology, and oncology—medical specialties that entail many of the CNIPA [11], EPO, JPO [12], KIPO [13], the United States
data-based tasks and components. A better understanding of Patent and Trademark Office [14], and provides computer
which medical specialties will be impacted by AI in the near translations to English for the CNIPA, JPO, and KIPO.
future might shed light on what guidelines, policies, or
frameworks to dedicate enough efforts to next. We performed a systematic search for the period between
January 1, 2012, and July 20, 2022. A query was made using
Also, while the number of peer-reviewed papers on AI’s role the following keywords: deep learning, machine learning, deep
in health care and medicine, relevant patents, and commercially neural networks, or artificial intelligence, in combination with
available AI and ML devices keeps on growing at an medical, medicine, healthcare, or health. In addition, the
unprecedented rate, it will become increasingly difficult for “computing arrangements based on specific computational
regulators and policy makers to keep up with the pace of models” (G06N) of the Cooperative Patent Classification (CPC)
innovation [4]. was used [15]. Patents are classified with at least 1 CPC code
There are numerous health care– and AI-related patents and the G06N is assigned when the invention relates to AI or
worldwide. Inventors and researchers can submit their patents ML techniques.
to national and international offices, of which the largest ones The following variables were extracted from the Espacenet
include the China National Intellectual Property Administration database: patent title and abstract, inventors, applicants,
(CNIPA), the European Patent Office (EPO), the Japan Patent publication number, CPC code, and publication date. As
Office (JPO), the Korean Intellectual Property Office (KIPO), inventors are allowed to submit their patent at multiple patent
or the United States Patent and Trademark Office. These 5 offices, duplicate patents were removed based on matching
largest patent offices collaborate in the Five IP Offices titles, inventor names, and applicant names. The patent
collaboration, making all their patents available in the Global publication number was used to identify the patent office that
Dossier initiative [5]. registered the patent.
Not every patent will lead to a product or a service on the Downloading Patent Abstracts
market, and even for those that succeed, it usually takes years
In total, 12,384 matches were found using the Espacenet search,
to reach the market and end up being a commercially available
based on which we performed the analysis. Search queries might
product or a product used in the medical practice.
contain overlapping results; therefore, we excluded duplications,
For example, a patent was submitted for “wireless transmission finally retaining 10,967 distinct matches.
of ECGs in handheld devices” in 1998 in the United States [6].
Public information was downloaded for all of the patents using
The applicants of the patent developed the idea of a smartphone
a Chrome-based crawler from Espacenet, followed by the
case that served as a single-lead electrocardiogram to be
extraction of titles and abstracts from the HTML source. The
approved by the FDA in 2012, a total of 14 years later. The
resulting text data were saved to files for further processing.
evolution of its design resulted in a credit card–sized device
Crawling was performed in August and September 2022.
and an even smaller version in 2021. In the meantime, the
company AliveCor received clearance by the FDA to use an Preprocessing of Textual Data
algorithm for the analysis of the readings to determine issues The first publication date and number were used wherever
related to cardiac rhythm without human intervention [7]. It multiple were available.
took around 2 decades for a digital health technology to
transition from a patent phase to becoming commercially We retained patents that were dated after January 1, 2016,
available, and years to build AI analysis into the device. excluding 197 (1.79%) patents of the available data set. The
last fully covered month was June 2022, the few patents (n=27,
It is pivotal to obtain a clear overview on upcoming AI and ML 0.25%) in July 2022 were excluded.
trends in health care to provide regulators with a better position
to foresee what technologies they will have to create regulations Some of the most frequent words of the English language were
for that are not available in the market yet. excluded from the analysis, as they would rank high in
appearance statistics without highlighting the trends we are
Therefore, in this study, we provide insights and forecasts into looking for. The excluded words were the following: “for,”
the trends in AI and ML in health care by conducting an “from,” “and,” “with,” “on,” “of,” “a,” “the,” “to,” “is,” “an,”
international patent analysis. “by,” “are,” “in,” “can,” “or,” “that,” and “be.” Additionally,
commas and parentheses were removed, and the text was
converted to lowercase.

https://ai.jmir.org/2023/1/e47283 JMIR AI 2023 | vol. 2 | e47283 | p. 2

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR AI Benjamens et al

Statistics Generated From Downloaded Data were introduced above, along with Query and Toplists pages.
Multiple statistics were generated from the patents; these were Besides these, the most frequent word pairs (eg, image
evaluated separately for titles and abstracts as well, except for segmentation) are also listed with the number of occurrences
top lists. separately for titles and abstracts. Finally, under “Trending,”
one can find those expressions whose occurrence rises steadily
Occurrence Counts: Single Words within the last 5 examined quarters, possibly highlighting
Titles of all used patents were merged into 1 string, and the methods that are currently becoming popular. The “Trending”
number of times each word appears was counted. The page examines 3 separate properties: change in absolute and
occurrence count list was constructed the same way for abstracts relative occurrence, along with the shape of the increase by
as well. Each word appearance was counted, not limited to 1 quarters correlated to a linearly increasing line in the (the
per patent. “Trend” column).

Abstract Query Interestingly, most of the single words with a high relative
increase are linked to modern technologies within health care
Some further cases were not covered by the abovementioned
(“device,” “forecasting,” “inference,” and “classifying”).
lists: expressions consisting of 2 or more words (eg, “brain ct
Similarly, some of the increasingly used word pairs are
image”), or words from the abstracts that are not listed above
“computer aided” and “learning algorithms.”
due to a very low occurrence count. A researcher could look up
arbitrary texts using the query form.
Results
After performing the preprocessing steps, the query string was
looked up in each patent’s title. The number of patents with By using the patent database filter option “Applicant toplist,”
matches was counted—that is, each patent is counted once at a list in descending order of the number of patent applications
maximum—as opposed to the “occurrence counts” described per applicant was generated. The number of patents applied by
above. an entity ranged from 1 to 305, with applicant Ping An
Technology (Shenzhen, China) filing for the highest number of
Additionally, appearance counts were displayed on a time scale patents (n=305) and several dozens of applicants filing for the
as well to visualize trends in 3-month units. lowest number of patents (n=1). We identified 5848 patents
Furthermore, to eliminate the effect of increasing patent count, with a company as the primary applicant and 3038 patents with
the relative frequency of the search term was also a university as the primary applicant. To derive insights relevant
displayed—this is useful to determine the trends of methods for the purposes of this study, the 20 applicants from this list,
because raw occurrence counts could increase even with a which applied for the most patents, were considered and the
declining technology when total patent counts increase over findings are summarized in Table 1.
time. Each entry in the “Applicant toplist” filter also lists the
Top Lists corresponding country, in abbreviated format, where the relevant
patent office is located. Out of the top 20 patent applicants, 14
Inventor, applicant, and CPC top lists are simple lists with are based in China, 3 are based in the United States, and 1 is
occurrence counts, based on patent properties without any pre- based in Germany, Japan, and the Netherlands, each.
or postprocessing steps.
From these data and extending to applicants beyond the top 20
A list of the top 20 medical specialties and related terms was ones, a list of the top 10 countries where most patents were
curated (Multimedia Appendix 1): anesthesiology, cardiology, applied from was curated. As Table 1 indicates, most of the
dentistry, dermatology, emergency medicine, gastroenterology, relevant patent applications were filed in China, followed by
gerontology, family medicine or primary care, internal medicine the United States. Among this list of top 10 countries, 4 are
(ie, infectiology, endocrinology, and nephrology), neurology, located in Asia, 4 are located in Europe, and 2 are located in
obstetrics and gynecology, oncology, ophthalmology, pathology, North America. Textbox 1 shows the top 10 countries from
pediatrics, psychiatry, pulmonology, radiology or nuclear where relevant patents were filed.
medicine, surgery, and urology [16].
By selecting the “Patent office stats” option from the database,
Open Access, Interactive Database the general trend in the number of health care patents between
We made our database open access, which is available on The 2016 and 2022 in selected patent offices was observed. There
Medical Futurist website [17]. The page allows visitors to were 156, 340, 747, 1552, 253, 4097, and 1278 AI- and
analyze the patent database to validate our findings and discover ML-related health care patents in 2016, 2017, 2018, 2019, 2020,
other trends. The code is available upon request. and July 2021, respectively; this indicates a general increase in
the application of such patents during that time period in the
Users can select from among the available functions in the left
patent offices in China, the United States, and South Korea,
sidebar, while the content for the chosen page appears on the
while the offices in Japan and Spain have experienced little to
right side.
no change in the volume of patents. Figure 1, generated from
In this web-based open access database, term frequency–inverse the database, plots the number of patents in the selected patent
document frequency is applied for the purpose of frequency offices over this time period.
scoring. Single-word occurrences within titles and abstracts

https://ai.jmir.org/2023/1/e47283 JMIR AI 2023 | vol. 2 | e47283 | p. 3

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR AI Benjamens et al

Table 1. Top 20 patent applicants.

Number Applicant name Occurrences, n Country
1 Ping An Technology (Shenzhen) Co Ltd 305 China
2 Siemens Healthcare GmbH 219 Germany
3 IBM Corp 217 United States
4 Koninklijke Philips N.V. 110 The Netherlands
5 Ping An Medical and Healthcare Management Co Ltd 105 China
6 Ping An International Smart City Technology Co Ltd 103 China
7 Tencent Technology Shenzhen Co Ltd 90 China
8 University of Electronic Science and Technology of China 82 China
9 Zhejiang University 79 China
10 Shandong University 59 China
11 Beijing University of Technology 57 China
12 Tsinghua University 57 China
13 Fudan University 50 China
14 Canon Medical Systems Corporation 47 Japan
15 Beijing Baidu Netcom Science Technology Co Ltd 46 China
16 Tianjin University 45 China
17 GE Precision Healthcare LLC 45 United States
18 Huazhong University of Science and Technology 45 China
19 Beihang University 44 China
20 General Electric 44 United States

Textbox 1. Top 10 countries from where patents were filed.

The top 10 countries from where patents were filed were as follows:
1. China
2. United States
3. South Korea
4. Germany
5. Japan
6. The Netherlands
7. Canada
8. India
9. United Kingdom
10. France

https://ai.jmir.org/2023/1/e47283 JMIR AI 2023 | vol. 2 | e47283 | p. 4

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR AI Benjamens et al

Figure 1. Patent trends in selected patent offices between 2016 and 2022. cn: China; ep: Spain; jp: Japan; kr: South Korea; us: United States.

The rate of increase in the number of patents varied in each trends showed a similar increase. The 4 most used terms were
office that experienced such an increase. A marked increase “cancer” (n=161), “tumor” (n=151), “radiotherapy” (n=55), and
was noticeable in China from mid-2017, around 2018 in the “malignant” (n=47).
United States, and only around 2020 in South Korea. The patent
When focusing on terms related to AI and ML, trends in AI-
office in China experienced a steady increase in the number of
and ML-based patents could be established. The 4 most used
applications with some notable dips in 2020, 2021, and 2022.
AI- and ML-based terms were “artificial Intelligence” (n=2450),
Despite those downtimes, that patent office maintained its lead
“neural network” (n=2043), “machine learning” (n=1717), and
during the time period analyzed.
“deep learning” (n=1492). An increase in the occurrence of
To analyze patent trends around medical specialties, we created these AI- and ML-based terms was identified between 2015 and
a database of words and expressions that are relevant to each 2021 (Figure 3).
of the major 20 medical specialties (Multimedia Appendix 1).
To demonstrate what kind of patents were included in the
When analyzing single words that appear in the title of patents, database, we chose to feature examples of recently registered
the top 5 medical specialties with the highest number of patents patents of the top 4 applicants: Ping An Group listed a patent
were radiology, oncology, cardiology, pulmonology, and surgery within the scope of the specialties of radiology and oncology,
with 394, 271, 128, 103, and 76 patents, respectively. titled “Lymph node metastasis prediction method and device,
equipment and storage medium” (CN113920137a) in January
The “Abstract - query” option of the database outputs the
2022. This patent focuses on the detection of lymph node
number of times the search term occurs in the abstracts. Using
metastasis in pancreatic ductal cancer on computed tomographic
the preselected specific terms for medical specialties, the
imaging of the abdomen. The results of the first clinical
occurrence of specialty-related terms was identified. Based on
application were published in January 2023 [18].
this list, the terms relating to radiology or nuclear medicine
occurred the most in abstracts (n=1160), followed by oncology Siemens Healthineers AG listed a patent within the scope of
(n=532), ophthalmology (n=454), surgery (n=309), pulmonology the specialties of radiology and pulmonology, titled “Assessment
(n=261), cardiology (n=252), and obstetrics and gynecology of abnormality patterns associated with covid-19 from x-ray
(n=217; Table 2). images” (US2022022818a) in January 2022. A full package of
AI solutions for COVID-19 imaging became commercially
When focusing on one of the medical specialties with a high
available the months thereafter [19].
number of patents (for instance, radiology), trends in
imaging-based patents could be established. The 8 most IBM Corp listed a patent within the scope of the specialties of
frequently used imaging-related terms were “image processing” pathology and oncology, titled “Interpretation of whole-slide
(n=682), “image data” (n=674), “imaging” (n=657), “image images in digital pathology” (US2022164946A1) in May 2022.
segmentation” (n=328), “CT image” (n=288), “X-ray” (n=120), The code, data, and models were published in January 2022 and
“MRI” (n=114), and “ultrasound” (n=77). An increase in the a Python-based package (for modeling and learning) is freely
occurrence of these imaging-related terms was identified available on GitHub [20,21].
between 2015 and 2021 (Figure 2). For the field of oncology,

https://ai.jmir.org/2023/1/e47283 JMIR AI 2023 | vol. 2 | e47283 | p. 5

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR AI Benjamens et al

Koninklijke Philips N.V. listed a patent within the scope of the Philips Cardiologs arrhythmias diagnostic software, which is
specialty of cardiology, titled “Systems and methods for commercially available and FDA-cleared under section 510(k)
identifying low clinical value telemetry cases” of the Food, Drug and Cosmetic Act [22].
(US2022020478A1) in January 2022. This patent is part of the

Table 2. Occurrence of specialty-related terms.

Number Specialty Occurrences, n
1 Anesthesiology 72
2 Dentistry 41
3 Cardiology 252
4 Dermatology 112
5 Emergency medicine 157
6 Gastroenterology 84
7 Gerontology 37
8 Family medicine or primary care 28
9 Internal medicine 174
10 Neurology 77
11 Obstetrics and gynecology 217
12 Oncology (ie, radiation oncology) 532
13 Ophthalmology 454
14 Pathology 87
15 Pediatrics 18
16 Psychiatry 94
17 Pulmonology 261
18 Radiology 1160
19 Surgery 309
20 Urology 28

Figure 2. Trends in imaging-based patents. CT: computed tomography; MRI: magnetic resonance imaging.

https://ai.jmir.org/2023/1/e47283 JMIR AI 2023 | vol. 2 | e47283 | p. 6

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR AI Benjamens et al

Figure 3. Trends in AI- and ML-based patents. AI: artificial intelligence; ML: machine learning.

entail more creative (vs repetitive) tasks might receive fewer

Discussion AI patents; thus, those could be less prone to AI- or ML-based
Principal Findings innovations [26].

Based on the identified health care–related AI-based patents, The discrepancy between the top-ranking medical specialties
China clearly stands out as a leader in the field of AI. Also, in the title and abstract analyses could be attributed to the higher
2020 seems to be a turning point with marked growth in health occurrence of related terms in the abstracts than in the titles,
care–related patents. Following the widespread success of given the higher density of words in the former.
ChatGPT (OpenAI) in 2022, there are no data to indicate that With this study, we attempt to prove the point that in the age
this growth would slow down [23]. of automation, preparing with regulations in time should be of
Certain medical specialties stand out in terms of the number of high priority among policy makers. The #wearenotwaiting
patents that have been submitted about AI technologies and movement that comprises thousands of patients with diabetes,
inventions that might be relevant to them (Figure 4). Based on who created artificial pancreatic systems, further emphasizes
a previous study published by our group about FDA-approved this [27]. These patients have developed applications, platforms,
AI- and ML-based medical technologies [1], radiology, and other solutions to help each other manage their diabetes.
cardiology, and oncology were already identified as specialties Their OpenAPS (Open Artificial Pancreas System) software
with many AI-based applications. The more repetitive or that was created entirely by the patient community with no
data-based tasks a specialty entails, the higher the potential for contribution from medical professionals automatically provides
automation to be able to contribute to that field. patients with the right doses of insulin based on their blood
glucose level [28].
Moreover, patents that include the analysis of medical images
or videos can be relevant to a range of specialties from radiology Due to the influx of advanced technologies such as wearable
to pulmonology and surgery. Specialties that are closely linked health sensors, portable diagnostic devices, and AI and ML
to medical imaging can also be in the focus of AI patents in the applications in health care, it has become inevitable to design
coming years. Examples include dentistry, ophthalmology, and regulations and guidelines for technologies that are not available
emergency medicine. in the market yet, but everything, including patent trends,
indicates that they will soon be. As patients now have access
Besides these imaging-oriented specialties, as analyzing images to technologies, data, and algorithms, they will find a way to
is a widely popular use case of AI and ML, dermatology and use the technology that is not yet regulated but can still help
pathology could also benefit from the AI revolution. In them manage their condition or health.
dermatology, the rise of skin-checking applications that can
analyze photos of skin lesions on patients’ smartphones The recent rise of the conversational agent and large language
underscores this observation [24]. In pathology, automated model ChatGPT and AI-based image generators such as
assessment of digitized histopathology slides falls into the same MidJourney and DALL-E all point toward this direction. As a
category [25]. response to ChatGPT, Google LLC published a study about
their own chatbot that was specifically designed to answer
Medical specialties such as psychiatry or neurology that are medical questions [29].
more interaction-based (as opposed to being data-based) and

https://ai.jmir.org/2023/1/e47283 JMIR AI 2023 | vol. 2 | e47283 | p. 7

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR AI Benjamens et al

We expect that by looking at medical and health care–related implementation into the practice of medicine and the delivery
AI and ML patent trends, regulators and policy makers could of health care.
better determine medical specialties, technological trends, or
A follow-up study that closely follows some of the patents and
areas such as imaging to dedicate more attention to. Thus, when
medical specialties that stood out in this analysis would be useful
a range of AI- and ML-based technologies become available in
to see and determine how much time it takes for an AI- or
those fields, proper regulations will ensure a safe and efficient
ML-based health care patent to reach the stage of practical
implementation.
Figure 4. The number of occurrences of specialty-related terms in healthcare AI patents assigned to each of the 20 medical specialties. AI: artificial
intelligence.

the literature about what keywords and expressions might be

Limitations associated with certain medical specialties, the database we
There are obvious limitations to our approach. As there is no generated is a subjective list of keyword-specialty associations.
globally accepted patent database, we could only focus on the Moreover, even if a specific medical specialty or its keyword
5 most active patent offices with the highest number of patents is mentioned in a patent’s abstract, it does not necessarily mean
worldwide. This implies that we might have overlooked patents that the patents are indeed associated with the specialty.
from other patent offices worldwide. As there is no database in

Authors' Contributions
SB, PD, and BM designed and conducted the study, GM designed the database, and all authors wrote the manuscript.

Conflicts of Interest
None declared.

Multimedia Appendix 1
Medical specialties and related terms.
[PDF File (Adobe PDF File), 92 KB-Multimedia Appendix 1]

References
1. Benjamens S, Dhunnoo P, Meskó B. The state of artificial intelligence-based FDA-approved medical devices and algorithms:
an online database. NPJ Digit Med 2020;3:118 [FREE Full text] [doi: 10.1038/s41746-020-00324-0] [Medline: 32984550]
2. Topol E. High-performance medicine: the convergence of human and artificial intelligence. Nat Med 2019 Jan;25(1):44-56
[doi: 10.1038/s41591-018-0300-7] [Medline: 30617339]

https://ai.jmir.org/2023/1/e47283 JMIR AI 2023 | vol. 2 | e47283 | p. 8

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR AI Benjamens et al

3. Artificial intelligence and machine learning in software as a medical device. US Food and Drug Administration. 2021.
URL: https://tinyurl.com/rwrh739a [accessed 2023-05-10]
4. Buiten M. Towards intelligent regulation of artificial intelligence. Eur J Risk Regul 2019 Apr 29;10(1):41-59 [FREE Full
text] [doi: 10.1017/err.2019.8]
5. Five IP Offices. URL: https://www.fiveipoffices.org/home [accessed 2023-05-10]
6. Buntz B. Ahead of his time. Medical Device and Diagnostic Industry. 2012. URL: https://www.mddionline.com/news/
ahead-his-time [accessed 2023-05-10]
7. Moynihan T. Ex-Googlers build a neural network to protect your heart. Wired. 2017. URL: https://www.wired.com/2017/
03/alivecor-kardia/ [accessed 2023-05-10]
8. Espacenet: free access to over 140 million patent documents. European Patent Office. URL: https://worldwide.espacenet.com/
patent/ [accessed 2023-05-10]
9. European Patent Office. URL: https://www.epo.org/ [accessed 2023-05-10]
10. Global Dossier and patent information (Working Group 2). Five IP Offices. URL: https://www.fiveipoffices.org/activities/
globaldossier [accessed 2023-05-10]
11. China National Intellectual Property Administration. URL: https://english.cnipa.gov.cn/ [accessed 2023-05-10]
12. Japan Patent Office. URL: https://www.jpo.go.jp/e/index.html [accessed 2023-05-10]
13. Korean Intellectual Property Office. URL: https://www.kipo.go.kr/en/ [accessed 2023-05-10]
14. United States Patent and Trademark Office. URL: https://www.uspto.gov/ [accessed 2023-05-10]
15. CPC scheme and CPC definitions. Cooperative Patent Classification. URL: https://www.cooperativepatentclassification.org/
cpcSchemeAndDefinitions/table [accessed 2023-05-10]
16. Association of American Medical Colleges. URL: https://www.aamc.org [accessed 2023-05-10]
17. The Medical Futurist. URL: http://www.medicalfuturist.com/patents [accessed 2023-05-11]
18. Bian Y, Zheng Z, Fang X, Jiang H, Zhu M, Yu J, et al. Artificial intelligence to predict lymph node metastasis at CT in
pancreatic ductal adenocarcinoma. Radiology 2023 Jan;306(1):160-169 [doi: 10.1148/radiol.220329] [Medline: 36066369]
19. AI COVID-19. Siemens Healthcare GmbH. URL: https://www.siemens-healthineers.com/medical-imaging/
digital-transformation-of-radiology/ai-covid-19-algorithm [accessed 2023-05-10]
20. Hierarchical Graph Representations in Digital Pathology. GitHub. URL: https://github.com/histocartography/hact-net
[accessed 2023-05-10]
21. Gibbs S. Artificial intelligence tool 'as good as experts' at detecting eye problems. The Guardian. 2018. URL: https://www.
moorfields.nhs.uk/sites/default/files/uploads/documents/Final [accessed 2022-11-09]
22. AI-enabled solutions. Philips. URL: https://www.philips.com/a-w/about/artificial-intelligence/
ai-enabled-solutions#triggername=less33_isc33 [accessed 2023-05-10]
23. van Dis EAM, Bollen J, Zuidema W, van Rooij R, Bockting CL. ChatGPT: five priorities for research. Nature 2023
Feb;614(7947):224-226 [FREE Full text] [doi: 10.1038/d41586-023-00288-7] [Medline: 36737653]
24. Novoa R, Gevaert O, Ko J. Marking the path toward artificial intelligence-based image classification in dermatology. JAMA
Dermatol 2019 Oct 01;155(10):1105-1106 [FREE Full text] [doi: 10.1001/jamadermatol.2019.1633] [Medline: 31411643]
25. Niazi M, Parwani AV, Gurcan M. Digital pathology and artificial intelligence. Lancet Oncol 2019 May;20(5):e253-e261
[FREE Full text] [doi: 10.1016/S1470-2045(19)30154-8] [Medline: 31044723]
26. Doraiswamy P, Blease C, Bodner K. Artificial intelligence and the future of psychiatry: insights from a global physician
survey. Artif Intell Med 2020 Jan;102:101753 [FREE Full text] [doi: 10.1016/j.artmed.2019.101753] [Medline: 31980092]
27. Wilmot E, Danne T. DIY artificial pancreas systems: the clinician perspective. Lancet Diabetes Endocrinol 2020
Mar;8(3):183-185 [FREE Full text] [doi: 10.1016/s2213-8587(19)30416-4]
28. Lewis D. Do-it-yourself artificial pancreas system and the OpenAPS movement. Endocrinol Metab Clin North Am 2020
Mar;49(1):203-213 [FREE Full text] [doi: 10.1016/j.ecl.2019.10.005] [Medline: 31980119]
29. Abduljawad M, Alsalmani A. Towards creating exotic remote sensing datasets using image generating AI. 2022 Presented
at: 2022 International Conference on Electrical and Computing Technologies and Applications (ICECTA); November
23-25, 2022; Ras Al Khaimah, United Arab Emirates [doi: 10.1109/icecta57148.2022.9990245]

Abbreviations
AI: artificial intelligence
CNIPA: China National Intellectual Property Administration
CPC: Cooperative Patent Classification
EPO: European Patent Office
FDA: US Food and Drug Administration
JPO: Japan Patent Office
KIPO: Korean Intellectual Property Office
ML: machine learning

https://ai.jmir.org/2023/1/e47283 JMIR AI 2023 | vol. 2 | e47283 | p. 9

(page number not for citation purposes)
XSL• FO
RenderX
 JMIR AI Benjamens et al

Edited by K El Emam; submitted 14.03.23; peer-reviewed by A Gao, JH Rajendran; comments to author 15.04.23; revised version
received 05.05.23; accepted 07.05.23; published 26.05.23
Please cite as:
Benjamens S, Dhunnoo P, Görög M, Mesko B
Forecasting Artificial Intelligence Trends in Health Care: Systematic International Patent Analysis
JMIR AI 2023;2:e47283
URL: https://ai.jmir.org/2023/1/e47283
doi: 10.2196/47283
PMID: 10449890

©Stan Benjamens, Pranavsingh Dhunnoo, Márton Görög, Bertalan Mesko. Originally published in JMIR AI (https://ai.jmir.org),
26.05.2023. This is an open-access article distributed under the terms of the Creative Commons Attribution License
(https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium,
provided the original work, first published in JMIR AI, is properly cited. The complete bibliographic information, a link to the
original publication on https://www.ai.jmir.org/, as well as this copyright and license information must be included.

https://ai.jmir.org/2023/1/e47283 JMIR AI 2023 | vol. 2 | e47283 | p. 10

(page number not for citation purposes)
XSL• FO
RenderX