1 s2.0 S0160791X23001264 Main

Technology in Society 74 (2023) 102321
Contents lists available at ScienceDirect
Technology in Society
journal homepage: www.elsevier.com/locate/techsoc
Artificial intelligence innovation in healthcare: Literature review,

exploratory analysis, and future research
Ahmed Zahlan *, Ravi Prakash Ranjan, David Hayes
Africa Business School, Mohammed VI Polytechnic University, Avenue Mohamed Ben Abdellah Regragui, Rabat, 10112, Morocco
A R T I C L E I N F O A B S T R A C T
Keywords: Artificial intelligence (AI) innovation in healthcare has emerged as an increasingly significant area of research.
Artificial intelligence AI, digital data collection, and computer infrastructure advancements have empowered humans to address
Innovation complex healthcare challenges. This study conducts a systematic literature review (SLR) of peer-reviewed journal
Healthcare
articles at the intersection of AI, innovation, and healthcare to offer research directions for scholars and leaders
Systematic review
Startup
in healthcare management. To achieve this, the systematic review identified and analyzed 378 published studies
on AI innovation in healthcare. Evaluating these publications based on inclusion and exclusion criteria yielded
75 studies ultimately selected for comprehensive analysis. This research adds to the scope of previous in
vestigations by aiming to 1) emphasize the most crucial AI-based healthcare applications, 2) explore challenges
associated with AI integration in healthcare, and 3) examine student adoption and incorporation of AI into
existing healthcare curricula. We also conducted an exploratory study of over 2700 AI-enabled healthcare
startups worldwide to supplement our literature review. The SLR reveals several gaps within the research scope
and proposes corresponding future research directions. These future research directions will assist researchers
and enable healthcare professionals to develop legislation that accelerates the adoption of AI solutions in
healthcare, ultimately enhancing public access to efficient and effective healthcare services.
1. Introduction reliable e-data to drive algorithm development [9]. In healthcare, data

resides in disparate patient medical databases such as electronic health
1.1. Background records (EHR), which may include critical administrative and clinical
data relevant to the patient, including demographics, radiology reports,
The emergence of artificial intelligence (AI) within healthcare holds past medical histories, vital signs, medications, laboratory data, immu
much promise. AI in healthcare helps to enhance patient diagnoses, nizations, progress notes, and insurance information [10]. AI can
improve prevention and treatment, increase cost-efficiency, and serve as revolutionize many care areas and administrative procedures for pro
a way to provide equitable access and treatment for all [1,2]. viders, payers, government entities, pharmaceutical companies, and
The term AI innovation refers to the creation of new knowledge, patients. This future possibility is due to the increasing availability of
tools, and ideas [3] using the simulation of human intelligence with electronically kept patient medical information and data sophistication
machines (e.g., AI, including machine learning, deep learning, neural within the field [11].
networks, and reinforcement learning) that are programmed to think AI currently equals and surpasses human capability outside health
like humans and mimic their actions [4]. AI innovation is crucial within care in speech recognition [12] and autonomous transportation [13].
healthcare to create new methods for facilitating and tackling healthcare Within the healthcare field, AI is now being used to help in areas such as
challenges and discovering new opportunities. automatic image analyses of skin cancer [14], surgery [15], oncology
In general, AI has been employed over the past 60 years for various [16], psychiatry [17], and clinical decision support [18]. With the
purposes within different fields [5–7] but has seen limited success in the health sector’s current challenges of lack of access to healthcare for
healthcare sector [8]. This limitation was due to the unavailability of everyone [19], human workforce shortages, high service costs [20], and
electronic data (e-data) access, as AI requires high-quality, precise, and slow adoption of information technology [21], AI adoption faces future
* Corresponding author.
E-mail addresses: ahmed.zahlan@um6p.ma (A. Zahlan), ravi.ranjan@um6p.ma (R.P. Ranjan), david.hayes@um6p.ma (D. Hayes).
https://doi.org/10.1016/j.techsoc.2023.102321
Received 24 April 2023; Received in revised form 4 July 2023; Accepted 4 July 2023
Available online 5 July 2023
0160-791X/© 2023 Elsevier Ltd. All rights reserved.
A. Zahlan et al. Technology in Society 74 (2023) 102321
barriers to enhancing the field’s quality [22]. • A better understanding of AI innovation in healthcare research
Researchers envision the future of AI in healthcare according to three findings: We present the most common themes associated with the
possibilities [18]. The first possibility foresees no need for doctors: AI topic, including the use of AI in healthcare studies, the adoption of AI
will diagnose all patients. The primary reason for this position is in healthcare by different stakeholders, and the challenges facing the
centered on cost savings but with comparable patient outcomes. A sec inclusion of AI in healthcare, including privacy, ethics, and trust.
ond possibility is that more patients can be analyzed with AI, with These findings address our first and second research questions.
doctors examining fewer patient cases, thus reducing the number of • The ecosystem factors for AI healthcare startups: We attempt to
people treated by doctors and decreasing healthcare system spending answer the third research question by augmenting our SLR findings
[23]. A final possibility is that AI will aid clinicians by helping doctors with an analysis of more than two thousand global AI healthcare
make better clinical choices, improving patient outcomes and cost startups, demonstrating the limited availability of external financing
efficiencies. and unequal geographical distribution of AI healthcare companies.
In this paper, we present the current research concerning a) the state The findings from this study lead us to present a future research
of AI innovation in healthcare, b) its adoption, c) the development and avenue of exploring the factors that impact the development and
growth of AI-based healthcare startups globally, and d) future research sustainability of AI innovation healthcare startups.
directions for AI-healthcare researchers and leaders and delve into the • Future directions for healthcare leaders and researchers: We outline
implications of our findings for global public health. By connecting the promises of the field and potential research questions from three
research with real-world applications, we aim to highlight the value of different perspectives. First, we aim to present future avenues for
our paper for policymakers, healthcare professionals, and industry researchers to contribute to the theoretical knowledge regarding AI
stakeholders. healthcare. Second, we detail opportunities for healthcare leaders,
investors, and stakeholders to act in integrating AI to improve effi
1.2. Research questions ciencies and costs. Third, to our knowledge, no existing studies have
established a connection between the growth of AI healthcare
Due to the rapid development and constant evolution of AI tech startups and the limited research available on these ventures. This
nologies [24], we aim to research the tools and trends used in AI in research adds insights into the issue and provides research directions
healthcare by examining the latest trends and algorithms applied in this for future investigations.
field, and we provide an updated and comprehensive understanding of
the state-of-the-art methods employed in AI-driven healthcare solutions. Overall, we hope the review assists stakeholders such as medical
This research is crucial for various stakeholders, including researchers, schools, clinics, engineers, and governments in designing a plan for
healthcare professionals, and policymakers, as it offers valuable insights enhancing AI innovation in healthcare for emerging markets. The rise of
into the most effective AI tools and methodologies that can enhance AI-based healthcare businesses will improve information availability,
patient care, streamline operations, and optimize resources; thus, we enhancing the recommended research areas.
present the first research question. The paper’s structure is outlined as follows: Section 2, Systematic
Review Methodology, elucidates the methodology employed for the
Research Question 1. What are the current AI techniques and algorithms
review and selection of papers. Section 3, Descriptive Results, provides a
employed in healthcare, and what potential trends can be observed in their
comprehensive bibliometric analysis of the chosen articles. Section 4,
application?
SLR Findings: State, Concepts, and Trends, expounds upon the state-of-
Due to the adoption challenges and resistance associated with AI in
the-art in artificial intelligence (AI) within healthcare and the limita
various fields, particularly in healthcare [25], it is crucial to investigate
tions concerning technology integration and adoption by various
these challenges and identify potential solutions. By gaining insights
stakeholders, and Section 5, AI Healthcare Startups: An Exploratory
into the obstacles faced during AI adoption, we can build well-founded
Analysis, fills a gap in the limited available studies regarding AI
recommendations for practitioners in the medical field who are
healthcare startups. Section 6, Discussion, delves into the present state
considering or currently implementing AI technologies in their organi
of AI healthcare startups globally while highlighting the scarcity of
zations. Consequently, we attempt to answer the following question.
research on AI medical startups and presents future research directions
Research Question 2. What are the main challenges in adopting and potential research questions. Lastly, Section 7, Conclusions, offers
and integrating AI in healthcare, and what potential solutions can final reflections.
address these challenges?
Startups play a significant role in innovation [26,27]. We want to 2. Systematic Review Methodology
explore how these startups contribute to AI healthcare innovation and
the current state of research about AI healthcare startups. This under To maintain the quality of our study, we implemented a two-step
standing is essential for founders, healthcare professionals, and policy approach. First, we adhered to the guidelines proposed by Kuckertz
makers, as it helps them grasp the AI healthcare landscape through the and Block [28] by addressing all the questions for conducting a litera
lens of startup activity. Thus, we explore this topic through the following ture review. The authors provided ten essential questions and criteria
question. that reviewers should consider when evaluating systematic literature
reviews. Consequently, we responded to the questions in Table 1 to
Research Question 3. What are the trends in the current landscape of
fulfill the SLR requirements.
AI healthcare startups, including their funding, focus areas, and
The second step of the review is illustrated in Fig. 2, where we per
geography?
formed a systematic process following the Preferred Reporting Items for
We wish to study these topics through a systematic evaluation of the
Systematic Reviews and Meta-Analyses (PRISMA) statement [29] to
relevant literature and select articles that address these questions.
select research articles for inclusion in this SLR. In the first stage of our
investigation, we looked through the keywords listed in Table 2. The
1.3. Contribution
search was conducted with the selected keywords in different digital
libraries, including PubMed and Scopus; we also conducted supple
This SLR contributes to the growing research interest in AI innova
mentary searches using Google Scholar. Pubmed was selected based on
tion in healthcare in three ways. It provides.
their notable frequency of research publications on AI in healthcare
[30]. Scopus encompasses a broader range of journals compared to
PubMed and Web of Science, offering faster citation analysis that covers
2
Table 1 criteria resulted in the elimination of 246 papers and the retention of
Step suggested to be followed by Kuckertz and Block [28] to present an SLR. 132 articles.
Question Answered Additional papers were not included in the review because they did
not directly answer our research questions, were still in press, or did not
“Does the SLR acknowledge previous The SLR acknowledges previous
literature reviews (including meta- literature reviews. provide a complete text for review. As a result, we narrowed our focus
analyses)?” and identified 69 papers remaining for inclusion in the review.
“Does the SLR have an explicit and well- All the research questions are cited in Lastly, additional articles [32–37] were considered for inclusion in
explained research question that is also section 1.2 the SLR using (Connectedpapers.com)3 tool (examples in Figs. 3 and 4).
well-justified?”
“Is the research question correctly The research terms are presented in
We evaluated these additional publications based on our selection
translated into a (relevant) set of Section 2 criteria. For instance, the visual representation of papers associated with
search terms?” [38] contains papers unrelated to answering our research questions;
“Is the database selection explained and Database selection is explained in hence, they were eliminated from our review (Fig. 3). In contrast [33],
based on a clear rationale?” Section 2
was connected directly to Ref. [39], which was included in our review
“Is the literature selection up to date?” The publications are up to date
(including 2023 publications) since it meets all inclusion requirements (Fig. 4). After applying
“Does the SLR clearly express its The exclusion and inclusion criteria are Connectedpapers.com to our review, we ended with 75 papers in the
inclusion and exclusion criteria?” explained in section 2, Fig. 2 research.
“Does the SLR include one or more Answered in section 3. Descriptive
overview tables characterizing each results
study in the sample?”
3. Descriptive Results
“Does the SLR employ a particular The aggregation method is explained in
aggregation and presentation Section 2 Although AI innovation in the healthcare industry is a relatively new
method?” topic in the scientific literature, the results indicate a growing trend in
“Does the SLR contribute beyond merely The SLR presented AI’s state of the Art in
the number of articles researched on this subject. Our paper selection
offering an analysis of the status quo of healthcare and went beyond meta-
the literature?” analysis. resulted in 75 papers, which indicates a noticeable increase in the
“Does the SLR provide implications for The SLR presents future research in number of studies published per year from 2017 onwards, with a peak in
(future) research and practice?” section 6 2019. While there is some fluctuation in the count for subsequent years,
the overall trend demonstrates a growing interest and research output in
the field (Fig. 5).
Table 2 The 75 papers appeared in 56 distinct journals. The journals that
Search string. include the most papers on the topic are Artificial Intelligence in Med
Digital String
icine (8 articles), BMC Health Services Research (3 articles), Npj Digital
Database Medicine (3 Articles), Intelligence-Based Medicine (2 articles), Journal
of the American Medical Association (2 articles), and Journal of the
Scopus ((Artificial AND intelligence) OR (Machine AND Learning)) AND
Innovation AND Health* American College of Surgeons (2 articles). The scientific journals that
PubMed Artificial Intelligence innovation in healthcare publish the most articles on the topic are often positioned in the top
quarter of their category (Table 3), indicating the importance and
relevance of these publications and the topic. Otherwise, the dominance
more articles than the citation analysis provided by other sources [31]. of medical journals in the table suggests that research in AI innovation
Our findings reveal crucial insights into publication trends and pat within healthcare has primarily been focused on the medical and clinical
terns. As illustrated in Fig. 1, the number of published papers increased aspects rather than business and management topics.
ten times from 2018 to2022,1 demonstrating an exponential increase in Within the literature, 67% are open access (Table 5); moreover, case
scholarly attraction to this emerging research field. The number of AI studies research predominates (29 out of 75 articles) the literature on AI
innovation in healthcare-related publications has increased exponen in healthcare, followed by review articles (22) and quantitative studies
tially since 2013 (n = 11X). It has nearly doubled in 2022 (n = 1136) (11), with the total number of articles per document type listed in
since 2021 (n = 737), as indicated by the statistical charts (Fig. 1). This Table 4. In terms of geographical distribution, the articles focus pri
growth trendline demonstrates that AI In healthcare has garnered sig marily on applications in the US (28 out of 75 Articles), the United
nificant attention from businesses and academics over the past ten years. Kingdom (10), Canada (6), and Germany (5); the complete list is pre
The initial screening of the outcomes started by employing keyword- sented below (Table 6).
based filters, targeting titles and abstracts that encompassed both “AI” or Among the articles reviewed, we observed that the most researched
“Artificial Intelligence” and “healthcare” keywords, which resulted in AI in healthcare concepts are the implementation of AI in healthcare, the
378 papers. Only article citations that met the following criteria were future of healthcare studies with AI, and privacy and ethical concerns
included: (a) peer-reviewed published journal articles classified in Q1 regarding applying AI in healthcare. Table 1 details the distribution of
journal based on (scimagojr.com)2 rather than working/conference pa concepts found in the SLR.
pers or unpublished material; (b) primarily focused on algorithms and As shown in Table 7, most of the selected publications focus on the
tools used in terms of AI in healthcare, focusing on challenges facing the use of AI in healthcare, AI healthcare studies in healthcare, and trust,
integration of AI in healthcare, AI in medical curriculum, and AI privacy, and ethics in the field of AI in healthcare. Additionally, the SLR
healthcare startups; and (c) paper written in English, these selection demonstrates a lack of research concerning AI-based healthcare ven
tures, identifying just three publications that study AI-based healthcare
startups.
1
The following presents findings from our literature review related to
This paper analyses literature through June 2023; we consider complete the most studied SLR concepts.
year-to-year comparisons in this observation, so the partial results for 2023 are
excluded.
2
Scimago Journal & Country Rank (Scimagojr.com) is a website that pro
vides information and metrics related to scientific journals and research in
3
stitutions. It offers a comprehensive ranking database of scientific journals Connectedpapers.com is an innovative visual tool designed to assist re
based on various factors, including citation metrics, scholarly output, and in searchers and applied scientists in locating and examining papers pertinent to
ternational collaboration. their field.
3
Fig. 1. Growth trend of Artificial Intelligence in healthcare literature for the last decade.
Fig. 2. Literature search and identification procedures.
4
Fig. 3. “Connectedpapers.com” tool for papers unrelated to the research [38].
Fig. 4. “Connectedpapers.com” tool for included papers related to the research [39].
4. SLR findings: state, concepts, and trends surgeries. Several areas of healthcare have been transformed by AI, as
outlined below.
In this section, we present details and findings from our SLR. First, Radiology has seen tremendous growth in the last few decades. Some
we summarize the concepts extracted from the literature review and AI (deep learning methods) allow radiologists to detect cancers using AI
then list the relevant applications of AI in different healthcare specialties successfully, such as by analyzing mammogram data sets with a detec
and diseases. Then, we outline the challenges to adopting AI in health tion rate of 0.90 [42]. In addition, researchers [82] have applied ma
care. Finally, we discuss AI adoption in healthcare education. chine learning techniques to magnetic resonance imaging (MRI),
enabling 98% accurate detection and segmentation of meningiomas
surrounding the brain and spinal cord.
4.1. State of the art - AI algorithms and their application in healthcare Cardiologists have used AI for coronary collapse early detection by
utilizing the gated recurrent unit (GRU) model, an extension of the
AI is transforming how physicians deal with patients and their con machine learning method. Additionally, researchers have been able to
ditions. AI brings various breakthroughs in the healthcare industry, predict impending heart failure in more than 88% of their research data,
including imaging, tracking, monitoring, and performing complex
5
Fig. 5. Number of publications per year in our literature.
thanks to data gathered by AI devices [87]. More conventionally, AI has AI in healthcare. These challenges include a need for more trust, privacy,
enabled researchers to create cardiologist-friendly wireless, remote MCT and ethics regarding the use of AI in healthcare and potential accuracy
(mobile cardiac telemetry) to capture, track, and relay various cardiac issues that arise when applying AI technology from the development
conditions of patients [99]. A study comparing MCT to Holter or Event perspective.
Monitors [100] reveals that employing MCT saves $141,811 on pro
cedures and surgeries compared to Holter monitors. Patients using MCT
4.2. Challenges for AI healthcare adoption
are better tracked (61% higher) than those using Holter or Event
Monitor.
Patients often opt for electronic devices to gather information about
AI shows high performance in oncology and outperforms pathologist
their healthcare [101]. Additionally, they believe that technology
diagnoses. For example, a team of researchers applied a neural network
should conform to established regulations and laws, uphold innovative
process (AI algorithm) to an extensive series of images; rapidly
principles, and safeguard freedom of expression to serve the general
analyzing the photos, the algorithm identified breast cancers with a 93%
public’s best interests. However, numerous obstacles hinder the wide
detection accuracy [39]. The neural networks method took only a few
spread integration of AI in healthcare. A significant concern revolves
seconds to accomplish the task, while a pathologist would take
around the potential unauthorized access to patients’ information, thus
approximately 30 h to achieve the same task. Another study utilizing
necessitating the resolution of such challenges [102]. Commercial
data from the National Cancer Registration and Analysis Service
business manipulation and behavioral marketing of personal data have
(NCRAS) in England found that machine learning accurately predicted
eroded patients’ confidence in how private and public organizations use
the diagnosis of lung and colorectal cancers with nearly 90% accuracy
their data [51]. This matter poses a significant concern for the patients
[16]. AI can also capture physical and mental health states; an AI-based
and impacts the adoption of the technology.
study predicted the development of psychosis with 100% accuracy in
The number of users who trust the privacy of AI data significantly
participants, a rate higher than clinician predictions [88].
impacts how much they rely on it [103]; most patients prefer a human
Interestingly, AI also predicts conditions before manifestation, such
physician rather than a robot doctor. To examine this preference, [57]
as type 2 diabetes and Celiac disease. Again, using machine learning
conducted several studies investigating patients’ healthcare acceptance
models on a “de-identified dataset,” researchers conducted a study that
of AI. In the first study, researchers studied whether respondents are less
yielded an 83% accuracy when diagnosing type 2 diabetes [41]. In a
likely to seek medical attention from automated services with equal
similar study, gastrologists utilized five AI techniques to create a clinical
performance with human assistance. Just around 27% signed up for the
decision-support system algorithm, trained on 178 clinical issues, to
automated services program, while 73% signed up for the human ser
predict Celiac disease [70]. This study yielded an eighty percent pre
vice. In a second study by the researchers, respondents indicated that
diction accuracy.
they would pay more for medical treatment by a human versus treat
AI has also been applied in dentistry to detect cavities and oral
ment from an AI-automated service, even if the automated service has
cancers and aid dentists. For instance, researchers trained complex
high accuracy.
neural networks (CNNs) on various dental pictures to detect cavities
To understand the reasons behind this resistance towards using AI in
from dental images [81]. They attained 88% and 89% diagnostic accu
healthcare, researchers conducted another study with 1421 re
racy for premolars and molars, respectively.
spondents, demonstrating that “individual” is a resistance to adopting
Based on the presented results, AI is progressively impacting the
AI. These findings are consistent with the belief that AI innovation ser
provision of healthcare services. Deep learning and neural networks
vices cannot account for individual characteristics or needs; respondents
have accelerated research in numerous fields, including biology. AI also
believe AI treats everyone the same way. Furthermore, the researchers
helps researchers treat crucial patients more successfully by increasing
suggest that this resistance could also be due to the affirmation of
their surveillance capabilities. These examples show that AI has started a
Promberger and Baron [104] that patients do not want to be accountable
new era in the disease control domain. Despite these successes, AI
for tough medical choices and therefore try to transfer blame to another
innovation in healthcare faces several obstacles, as outlined in the next
individual. Because humans are shown to take responsibility while
section. Our SLR finds that challenges exist concerning the adoption of
computers do not, patients are more likely to adopt a doctor’s advice
6
Table 3 Table 3 (continued )

Main sources of publications. Journal N. of Quartile Category
Journal N. of Quartile Category paper
paper
Globalization and Health 1 Q1 Medicine, Health Policy,
Artificial Intelligence in 8 Q1 Artificial Intelligence in Public Health
Medicine Medicine Information Fusion 1 Q1 Computer Science
BMC Health Services 3 Q1 Health Policy and Digital Academic Radiology 1 Q1 Medicine, Radiology,
Research Health Nuclear Medicine and
npj Digital Medicine 3 Q1 Medicine, Health Imaging
Informatics, Health npj Schizophrenia 1 Q1 Medicine, Psychiatry and
Information Management Mental Health
Intelligence-Based Medicine 2 Q1 Artificial Intelligence in Journal of the American 1 Q1 Medicine, Health
Medicine Medical Informatics Informatics
Journal of the American 2 Q1 Medicine International Journal of 1 Q1 Decision Sciences, Computer
Medical Association Information Management Science, Business,
Journal of the American 2 Q1 Surgery Management and
College of Surgeons Accounting
Nature Medicine 2 Q1 Medicine Journal of the American 1 Q1 Medicine, Radiology,
European Radiology 2 Q1 Medicine, Radiology College of Radiology Nuclear Medicine and
Journal of Medical Internet 2 Q1 Health Informatics Imaging
Research Journal of clinical medicine 1 Q1 Medicine
Frontiers in Medicine 2 Q1 Medicine Biochimica et Biophysica 1 Q1 Biochemistry, Genetics and
Jmir Medical Education 2 Q1 Education Acta Molecular Biology
Science Translational 1 Q1 Medicine Expert Systems with 1 Q1 Computer Science, Artificial
Medicine Applications Intelligence
Journal of Dentistry 1 Q1 Dentistry California Management 1 Q1 Business, Management and
Academic Medicine 1 Q1 Education, Medicine Review Accounting
Annals of Oncology 1 Q1 Medicine, Oncology BMC Medical Ethics 1 Q1 Medicine, Health Policy,
Annals of Emergency 1 Q1 Medicine Ethics and legal aspects
Medicine The international journal of 1 Q1 Engineering
Suicide and Life-Threatening 1 Q1 Medicine, Psychiatry and science
Behavior Mental Health Scientific reports 1 Q1 Multidisciplinary
BMC Medical Informatics and 1 Q1 Health Policy, Health International journal of 1 Q1 Medicine, Health
Decision Making Informatics medical informatics Informatics
Plos Medicine 1 Q1 Medicine, Bio (Chemistry, BMC Psychiatry 1 Q1 Medicine Psychiatry and
Technology, Biology) Mental Health
Plos One 1 Q1 Medicine
Insights into Imaging 1 Q1 Radiology, Nuclear
Medicine and Imaging
Nature Review Cancer 1 Q1 Medicine, Oncology, Cancer Table 4
Research Number of publications by paper type.
BMJ Global Health 1 Q1 Medicine, Health Policy,
Document Type Number of Publications
Public Health
Nature Biomedical 1 Q1 Medicine, Biomedical Case Study 29
Engineering Engineering, Review article 22
Nature Machine Intelligence 1 Q1 Computer Science Quantitative study 12
Medical Education Online 1 Q1 Medicine, Social Sciences Qualitative Study 5
Frontiers in Psychiatry 1 Q1 Medicine, Psychiatry and Systematic Literature Review 4
Mental Health Opinion Article 3
Technology in Society 1 Q1 Business, Management and
Accounting
European Heart Journal 1 Q1 Medicine, Cardiology and
Cardiovascular Medicine Table 5
Stroke and Vascular 1 Q1 Medicine, Cardiology and Number of publications per access.
Neurology Cardiovascular, Neurology
Access N. Publications %
Technovation 1 Q1 Business, Management and
Accounting Open Access 50 66.67
Emerging Science Journal 1 Q1 Multidisciplinary Closed (Access Only through institution) 24 32.00
British Journal of 1 Q1 Medicine Short-term open access 1 1.33
Dermatology
Journal of the Association for 1 Q1 Decision Sciences
Information Science and Information Systems and than the recommendation of a robot.
Technology Management
Another study [105] shows that 63% of adults are dissatisfied with
PharmacoEconomics 1 Q1 Medicine, Health Policy,
Pharmacology having their data used to improve healthcare and are against using AI in
Journal of Consumer 1 Q1 Business, Management and their traditional roles to replace actual doctors and nurses. Further
Research Accounting, Economics research explored in the US by Intel Corporation [106] shows that even
Heart: Cardiology in focus 1 Q1 Cardiology and
though the attitudes to AI in healthcare organizations across the country
Cardiovascular Medicine
European Journal of 1 Q1 Business, Management and
are positive regarding implementation, half of the participants think it
Innovation Management Accounting will result in fatal errors and will not live up to expectations. These
International Journal of 1 Q1 Business, Management and survey results provide insight into the challenges of using AI in health
Entrepreneurial Behavior Accounting care, particularly data ownership, privacy and fairness, and liability
& Research
issues.
British Journal of 1 Q1 Medicine
Haematology There is also a possibility AI system trained on a small amount of
inadequate or skewed data from different sources may have incorrect
results, of which a consumer might need to be made aware. We can see
7
Table 6 years of medical school, general training, internships, and a time of

Number of publications per country. specialization. While modern medicine has seen many advancements,
Country of Publication Number of Publications medical education continues to adhere to classic teaching methods
[109]. Medical students spend considerable time in medical school,
USA 28
United Kingdom 10 accumulating as much experience as possible and then learning how to
Canada 6 apply it to their patients. As a result, there is a high reliance on prior
Germany 5 knowledge in this approach.
The Netherlands 4 Many physicians, nurses, pharmacists, and other healthcare pro
Italy 4
Brazil 2
viders argue that modern medical science education should move
Switzerland 2 beyond pure theory to include education and training in AI technologies
India 1 [110]. To do so, researchers suggest how AI can be integrated into
Ireland 1 medical studies [45]; the authors propose introducing linear algebra,
Greece 1
calculus, and probability into the Medical College Admission Test
Japan 1
Spain 1 (MCAT). In addition, they assert that medical students should consider
Sri Lanka 1 working with medical data sets, EHR, and AI during the core phase of
Singapore 1 medical studies. The students should receive training in the fundamental
Hungary 1 concepts of data and AI-based clinical applications. Others emphasize
Saudi Arabia 1
Sweden 1
the need for medical schools to work closely with colleges and de
Australia 1 partments specializing in engineering and computer science subjects
China 1 [111].
South Korea 1 Another significant medical technology advancement that receives
Israel 1
less publicity is using electronic health records (EHRs); in many devel
oped countries, this still needs to be routinely taught to medical students
this in Ref. [84], who trained a machine learning model to identify skin and clinicians [45]. Medical students should have explicit knowledge of
cancer with only 5% of the data from African-American people. The the 4 V’s (volume, variety, velocity, and veracity) of big data and a
results could be biased toward African Americans. Understanding these complete grasp of how these principles apply to an action in the equa
factors creates uncertainty in the patient’s mind, lowering the confi tion of medical field decisions [112]. The researchers also assert that
dence to accept AI-based solutions. medical students should be familiar with both general and unique uses
This finding poses an important question: How can we enhance patient of AI principles.
trust in AI systems for healthcare? In a recent study conducted with 484 medical students from 19 UK
In an interview, Pat Baird, a senior regulatory specialist at the medical universities [50], 88% of the respondents believe that AI can
Phillips Corporation [107], mentions that enhancing patient trust in AI play a significant role in healthcare in the future. Moreover, 88% of the
in healthcare requires action on three levels. First, to achieve technical students agreed that instruction in AI would be advantageous for their
confidence by questioning if the Algorithm is doing what it is meant to future. Most students (78.1%) stated that all medical students should be
do. The second level involves regulatory trust, where technology must trained in AI as part of their medical degree. In another study that sur
meet the needs and standards of various users. The final level involves veyed undergraduate German medical students [32], 77.2% of re
human confidence, perhaps the most critical factor. Data and analyses spondents predicted that AI would revolutionize imaging and medical
must be derived by examining the heterogeneity of patient groups. Also, imaging. Furthermore, 83.7% agreed that doctors would use AI to di
reliable protection against cyberattacks must be provided to ensure agnose abnormalities from digital imaging. A similar study in Canada
leaks and misuse of patient data. Also, audit trails should be adequately [49] focused on the effect of AI on medical students’ choice of radiology.
maintained to ensure that who does what is logged, what documents are Among all the 322 respondents, 78.9% understand AI well, and 94.7%
used are recorded, and what system modifications are made are known agree that AI would increase radiologist capacity and efficiency.
[51]. Another qualitative study [37] shows a strong consensus on the appli
There is a particular urgency to establish a comprehensive under cation of AI in administrative environments and research involving
standing of trust between artificial intelligence and humans in the health-related data. However, students voiced concerns about the po
healthcare sector. Many machine learning algorithms in healthcare are tential insufficiency of data protection and the prospect of increased
subject to privacy and ethical constraints, forcing developers to take workplace surveillance in the future. The assessments suggested that
extra care when embarking on data collection and processing [53]. upcoming physicians are eager to engage more deeply with AI in the
Recently, the Food and Drug Administration (FDA) solicited public input medical field.
and issued guidelines for governing AI systems that perform data up From these studies, we see a clear awareness among future physi
dates automatically and on how to deal with updatable algorithms cians of the value of AI in healthcare. Students feel that artificial intel
[108]. ligence does not pose a threat because, in the end, there will be no
This discussion shows that it is necessary to consider a few key fac rivalry between humans and AI. Furthermore, these future doctors
tors when dealing with AI in healthcare. First, data accuracy should be embrace the prospect of working with AI to simplify patients’ lives.
evaluated and deemed reliable by cross-checking its veracity from other A key question arises from these conclusions: How can we integrate AI
sources or users. Second, ethical guidelines for handling sensitive into the medical curriculum?
medical information must be followed throughout deliberations on While exploring this question, we observe that universities world
implementing AI systems in healthcare. Third, there is a need for a wide have proposed several initiatives to integrate AI into the medical
structure to build trust between humans and machines. curriculum. Carle Illinois College of Medicine collaborates with physi
Furthermore, our SLR reveals difficulties surrounding AI adoption in cians, engineers, practitioners, and technical scientists to present cour
healthcare education, which will be elaborated on in the next section. ses about new medical technologies [113], and Stanford University
Center for AI in Medicine and Imaging works on collaborative projects
involving graduates and postgraduate students to solve health chal
4.3. AI adoption in medical education lenges using machine learning [114]. Furthermore, the University of
Virginia Center for Engineering in Medicine provides a program where
Before becoming a doctor, a physician must undergo several clinical medical students collaborate with engineering labs to develop novel
8
Table 7
SLR Concept matrix for AI innovation in healthcare.
Concepts
Publication
(A) Artificial (B) (C) Specialty (D) (E) Disease (F) Clinical (G) AI (H) (I) (J) (K)
intelligence Healthcare (Radiology Machine (Diabetes, decision Healthcare Healthcare Ethics Privacy Trust
et al.) Learning Cancer, DSM support startups Education
…)
[1] ✓
[5] ✓ ✓ ✓
[12] ✓ ✓ ✓
[14] ✓ ✓
[15] ✓ ✓
[16] ✓
[17] ✓ ✓
[18] ✓ ✓
[32] ✓ ✓
[33] ✓ ✓ ✓
[34] ✓ ✓ ✓
[35] ✓ ✓ ✓
[36] ✓ ✓
[37] ✓ ✓
[38] ✓ ✓ ✓
[39] ✓ ✓
[40] ✓ ✓ ✓
[41] ✓ ✓
[42] ✓ ✓
[43] ✓ ✓
[44] ✓ ✓ ✓
[45] ✓ ✓ ✓
[46] ✓ ✓
[47] ✓ ✓
[48] ✓ ✓ ✓
[49] ✓ ✓ ✓
[50] ✓ ✓ ✓
[51] ✓ ✓ ✓ ✓
[52] ✓ ✓ ✓
[53] ✓ ✓ ✓ ✓
[54] ✓ ✓
[55] ✓ ✓
[56] ✓ ✓ ✓
[57] ✓ ✓ ✓ ✓ ✓
[58] ✓ ✓
[59] ✓ ✓ ✓ ✓
[60] ✓ ✓ ✓ ✓
[61] ✓ ✓
[62] ✓ ✓
[63] ✓ ✓ ✓
[64] ✓ ✓ ✓ ✓
[65] ✓ ✓
[66] ✓ ✓ ✓ ✓
[67] ✓ ✓ ✓
[68] ✓ ✓
[69] ✓ ✓
[70] ✓ ✓ ✓
[71] ✓ ✓
[72] ✓ ✓
[73] ✓ ✓
[74] ✓ ✓ ✓
[75] ✓ ✓
[76] ✓ ✓
[77] ✓ ✓
[78] ✓ ✓
[79] ✓ ✓ ✓
[80] ✓ ✓ ✓
[81] ✓ ✓ ✓
[82] ✓ ✓
[83] Ć ✓
[84] ✓ ✓
[85] ✓ ✓ ✓
[86] ✓ ✓
[87] ✓ ✓
[88] ✓ ✓
[89] ✓ ✓ ✓
[90] ✓ ✓
[91] ✓ ✓
[92] ✓ ✓ ✓
[93] ✓ ✓
(continued on next page)
9
Table 7 (continued )
Concepts
Publication
(A) Artificial (B) (C) Specialty (D) (E) Disease (F) Clinical (G) AI (H) (I) (J) (K)
intelligence Healthcare (Radiology Machine (Diabetes, decision Healthcare Healthcare Ethics Privacy Trust
et al.) Learning Cancer, DSM support startups Education
…)
[94] ✓ ✓
[95] ✓ ✓ ✓
[96] ✓ ✓ ✓
[97] ✓ ✓
[98] ✓ ✓ ✓
ideas [45]. diagnostics and early detection. As an overview of these startups, we

Lastly, Table 7 shows that only three papers included in the SLR present the following examples.
focusing on AI-Based startups in the medical field were from the Q1 Ibex is the leading AI company in pathology; a Galen image recog
quantile of journals [38,95,98]. The mentioned papers do not directly nition platform uses AI-based algorithms to evaluate and classify cancer,
address our research questions or provide insights into AI healthcare enabling pathologists to achieve diagnostic accuracy while saving time.
startups’ current state, distribution, and impact on innovation in SkinVision inspects skin lesions using the phone’s camera and examines
healthcare using artificial intelligence. To understand this subject skin cancer incidence.
comprehensively, we performed a supplement analysis to the SLR that Medical imaging firms use visual data to find and identify abnor
explores AI innovation in healthcare startups. The following section malities in images. Viz. ai helps doctors to use machine learning to
presents our secondary data analysis of AI healthcare startups, providing recognize abnormalities in brain scans. At the same time, See-Mode, a
insights into ecosystem factors that drive AI innovation in healthcare Singapore-based company, integrates medical imaging and AI with
startups. computational fluid dynamics to help physicians predict strokes,
allowing them to save lives.
5. AI healthcare startups: an exploratory analysis We also observe that several prominent corporations are making
investments in genetic research. For example, Lifebit is the world’s first
To address the identified research gap concerning AI healthcare life-genomics platform to comprehend and create personal conclusions
startups, we analyzed a dataset containing companies from around the about DNA data as humans do. FDNA, with USD35 M in total funding
globe. These startups implemented AI technology across various raised, AI technology enables the company to identify abnormal patterns
healthcare sectors, such as drug discovery, mental health, chatbot of physiological changes that lead to disease.
assistance, imaging, genetics, nutrition, hospital decision support, Fig. 6 displays the number of startups founded annually, illustrating
fertility, diagnostics, and pathology. We aim to understand AI technol the exponential growth in AI healthcare startups each year and high
ogy’s diverse applications and potential implications within the lighting a fascinating field garnering interest. The figure showcases the
healthcare sector. number of AI healthcare startups established annually from 2000 to
To perform this investigation, we gathered data from two central 2022, highlighting intriguing trends and patterns. In the initial years,
databases: Crunchbase [115] and CB Insights [116]. CB Insights is a from 2000 to 2002, only a few startups emerged, indicating the nascent
technology market intelligence platform that examines data on venture stage of the AI healthcare sector with limited interest and investment.
capital, startups, patents, and more to forecast market trends. On the However, a substantial upswing was observed from 2007 onward, with
other hand, Crunchbase specializes in providing private company pro the count increasing from 12 in 2007 to 116 in 2013, signifying rapid
specting and research solutions. expansion driven by advancements in AI technology and its potential for
A comprehensive data cleaning procedure was undertaken after the transforming healthcare. From 2014 to 2018, the subsequent years saw a
data collection process. Firstly, duplicate entries were meticulously consistent upward trend, rising from 139 in 2014 to 395 in 2018,
identified and eliminated to ensure the dataset comprised unique ob reflecting growing interest in AI in healthcare. However, starting from
servations. Secondly, we excluded any entries needing descriptions and 2020, there was a noticeable decline, with the count dropping from 289
countries, as such omissions could undermine the accuracy and reli in 2020 to 62 in 2022, potentially due to Covid-19, economic conditions,
ability of subsequent analyses. Finally, a rigorous screening exercise was shifts in investor priorities, or industry consolidation.
conducted to eliminate companies whose descriptions did not directly Table 9a,b offers a comprehensive overview of the distribution of AI
pertain to the domain of artificial intelligence (AI) in healthcare, thus healthcare startups across different countries. One key observation is the
ensuring that the resultant dataset was optimally suited for investigating significant concentration of AI healthcare startups in the United States,
research questions within this area of interest. Following this thorough the United Kingdom, and China, which dominate the global landscape
data-cleaning process, we arrived at a final dataset of 2747 AI-based together. The United States leads with an impressive 1197 startups,
healthcare startups worldwide. This comprehensive analysis promises followed by the United Kingdom at 177 and China with 145. This con
to provide valuable insights into the landscape of AI-driven healthcare centration may be attributed to a robust technology and innovation
startups and their potential to revolutionize the industry. Table 8 pro ecosystem, access to skilled talent, and government support within these
vides a summary of the additional exploratory data. countries. Moreover, we can see this reflected in funding, where the
Our analysis of these startups indicates that the largest category is United States leads the investment table with a significant amount of
represented by pathology and diagnostics startups that use AI for funding at $32.5 billion, suggesting a well-developed and well-funded
AI healthcare startup ecosystem in the country.
In comparison, China and the United Kingdom hold the second and
Table 8
Summary of AI medical startups database.
third positions in investments, with $7.1 billion and $2.9 billion,
respectively. Despite being lower than the United States, these figures
Total number of startups 2747
emphasize active investments in AI healthcare startups within these
Number of startups with funding data available 2211 countries. Table 9–a reveals a broader regional distribution, with North
Total fundraising (USD) 50.2 billion
America, Europe, and Asia hosting most AI healthcare startups. Nations
Number of Distinct Country 79
10
Fig. 6. Number of AI healthcare startups founded per year.
like Canada, Germany, France, Spain, Israel, India, South Korea, and
Table 9a
Japan contribute significantly to this industry. Interestingly, smaller
Top 20 countries in the number and funding amount for AI healthcare startups
countries such as Israel, Singapore, and Switzerland have dispropor
Country Number of AI HC Startups Funding (Million USD) tionately high numbers of startups compared to their population size.
United States 1197 32569.98 Emerging markets like Brazil, India, and South Africa show potential for
United Kingdom 177 2901.32 future growth. At the same time, regions like Africa, Central America,
China 145 7148.81 and South America are underrepresented, signaling untapped potential
Israel 131 1647.39
Canada 130 498.80
and opportunities for investment in AI healthcare initiatives (Table 9–b).
India 122 765.77 The discrepancies in funding accessibility among diverse global regions
South Korea 84 631.33 are noteworthy and adhere to similar patterns identified in other
France 74 758.85 research focused on artificial intelligence investment [117].
Germany 59 313.91
Examining the total funding aspect, Table 10 highlights a generally
Spain 55 242.47
Singapore 52 355.47 upward trajectory in financial support for AI healthcare startups. A
Japan 48 609.04 notable surge occurred between 2017 and 2018, nearly doubling the
Switzerland 41 330.84 previous year’s amount, reflecting heightened investor enthusiasm for
Australia 41 260.05 AI in healthcare. 2021 marks a significant milestone, with total funding
Taiwan 31 41.31
Netherlands 30 42.73
reaching $12.2 billion, indicating robust growth and market confidence.
Brazil 24 9.67 However, a decline is observed in 2022, with both deal count and total
Italy 22 25.03 funding decreasing. The exact reasons for this reduction are unclear.
Sweden 22 45.37 However, they may be attributed to market saturation, economic fluc
Hong Kong 18 446.42
tuations, challenges specific to the AI healthcare sector, or a post-COVID
effect. Monitoring this trend will be crucial to determine if it represents a
temporary shift in the investment landscape.
Table9b Our analysis of global AI healthcare startups reveals a rapidly
The least 20 countries on the number and funding amount for AI healthcare growing industry with diverse applications across various sectors. The
startups. United States, the United Kingdom, and China dominate the landscape,
Country Number of AI HC Startups Funding (Million USD) reflecting well-developed ecosystems and significant investments. While
Malaysia 2 47.60 smaller countries and emerging markets show potential for future
Vietnam 1 0.00 growth, regions such as Africa, Central America, and South America
Jordan 1 0.00 remain underrepresented, indicating untapped opportunities.
Cayman Islands 1 0.00
Peru 1 0.00
Belize 1 0.01 Table 10
Iceland 1 0.79 Amount of investment and deals per year for AI healthcare startups worldwide
Ivory Coast 1 0.04
from CB Insight [116].
Zimbabwe 1 0.03
Croatia 1 0.09 Year Deal Total Funding (Million USD)
Morocco 1 0.30 Count
Malta 1 0.00
2014 113 $1269.68
Georgia 1 0.00
2015 124 $1468.25
Kazakhstan 1 0.01
2016 198 $2010.25
Tunisia 1 0.00
2017 260 $3836.21
Slovenia 1 0.95
2018 337 $7701.88
Qatar 1 0.95
2019 373 $5383.27
Zambia 1 0.02
2020 441 $7154.19
British Virgin Islands 1 2.00
2021 511 $12,194.19
Mali 1 1.70
2022 482 $5923.97
2023 131 $1408.96
11
Furthermore, an upward trajectory in funding demonstrates the mar the causal factors behind investment constraints in AI healthcare inno
ket’s confidence in AI’s transformative potential for healthcare. Moni vation in these regions. This focus will contribute to a better under
toring the investment landscape and ongoing trends will be crucial in standing of the challenges and potential opportunities for AI innovation
understanding the future direction of this dynamic sector. in healthcare across diverse economic contexts.
The limited number of business management journals in the table
6. Discussion might indicate a need for greater interdisciplinary collaboration be
tween healthcare, AI innovation, and business management. Integrating
6.1. Observed research gaps business and management perspectives into AI-driven healthcare inno
vation can be crucial for addressing issues related to implementation,
About the selected journals, Table 3 demonstrates that out of 56 scalability, adoption, ethical considerations, and financial impact.
journals, only seven are related to business and management; the pre Additionally, to address the geographical gap for limited research in
ponderance of publications in medical journals suggests that the LMCIs, we encourage international collaboration and knowledge
research has primarily been focused on the medical and clinical aspects. sharing between researchers from high-income countries and LMICs. We
This disparity indicates a gap in the literature regarding the intersection promote inclusivity and diversity in AI research by actively seeking
of AI innovation in healthcare with business and management. Inter partnerships with researchers from LMICs and involving them in global
estingly, Young [98], a commentary paper addressing business man AI initiatives.
agement challenges faced by AI healthcare startups, was published in a This SLR shows the importance of gaining a more comprehensive
medical journal rather than a business or technology-focused understanding of the missing elements to drive healthcare innovation
publication. with AI. In a study [120] exploring innovation drivers in the United
Similarly, as detailed in Table 6, there is a lack of high-impact journal Kingdom, the author identifies shortening product life cycles, technol
publications from researchers in low-medium-income countries ogy, increasing global competition, and changing customer need as the
(LMICs). Notably, the USA and United Kingdom researchers authored sector’s primary innovation drivers in the private sector. The author also
half of the selected publications used in this SLR. This observation suggests that the critical drives for public firms to innovate include the
highlights a disparity in the representation of academic contributions growing demand for municipal services and fiscal constraints brought
and perspectives from various global regions. about by the recession. Those innovation drivers remain limited to a
Furthermore, the included exploratory study of 2747 startups un specific country and general for all domains.
covers an important research observation; despite the considerable Moreover, the SLR concludes that acceptance of AI in healthcare,
development and expansion of AI healthcare startups globally, there trust, a well-developed infrastructure for AI, and investment are essen
appears to be a relatively small number of research publications tial drivers for AI healthcare innovation. We assert that these drivers
explicitly addressing this subject. As seen in Table 7, only three publi need to be explored in more depth to understand better the impacts and
cations pertain to AI healthcare startups, signifying limited research in mechanisms of AI innovation in healthcare.
this domain. This finding is unexpected, particularly given the sub Another avenue for future research is understanding the acceptance
stantial market capitalization and investor investments in the field. The and adoption of AI innovation in healthcare by patients, physicians, and
lack of research in this area raises crucial questions, suggesting a po medical students, especially in emergent and low and medium-income
tential need for increased focus and investment of time to study and countries. Al Badi et al. [121] studied the challenges of AI adoption in
comprehend the impact, challenges, and opportunities presented by AI healthcare in the United Arab Emirates context. The challenges found by
healthcare startups. Expanding research in this field may offer valuable the authors include 1) accuracy, as the results need to be accurate and
insights into the efficacy, ethical considerations, and possible risks avoid any partial data; 2) privacy and security; and 3) personal
associated with incorporating AI technology into healthcare. healthcare record confidentiality. In future studies, researchers can
As evidenced by the table, the sparse number of research publica explore strategies for addressing the challenges associated with adopting
tions centered on AI healthcare startups emphasizes an underexplored AI in healthcare, particularly those highlighted in this literature. By
facet within the larger context of AI in healthcare. examining these obstacles, scholars can contribute valuable insights and
develop practical solutions for successfully integrating AI healthcare
6.2. Future research directions technology.
Moreover, Haider [122] presented the barriers to adopting AI
This systematic literature review study and the supplementary healthcare in India; the considerable expense, the initial investment, and
analysis presented in this paper indicate an increasing field of interest the necessary infrastructure to use AI in healthcare are significant bar
with considerable future potential for researchers and healthcare riers. In addition, new technologies, especially those involving artificial
leaders. Given the identified gaps, expanding research on the topic in intelligence, are often perceived as a threat by older people because of
several areas is crucial. their inability to trust and trustworthiness issues. Moreover, there is a
First, researchers should focus on identifying the characteristics of AI significant barrier to adopting AI solutions in healthcare due to an
healthcare startups to comprehend their role in expediting innovation incomplete mechanism for ensuring privacy, protection, consistency,
and their impact on potential healthcare system savings. The findings and accuracy. Future research could focus on developing an acceptance
indicate a necessity for heightened research efforts to bridge the gap model for AI in healthcare by either developing new theories or
between AI healthcare startup growth and academic comprehension of extending existing models, such as the Technology Acceptance Model
their implications concerning specialty areas, diseases, clinical decision (TAM) [123] and the Unified Theory of Acceptance and Use of Tech
support, and other essential aspects such as ethics, privacy, and trust. nology (UTAUT) [124], as suggested previously in related studies [125,
Researchers might also explore the distribution of high investment and 126].
the prevalence of AI healthcare startups in the USA, considering that the Finally, as medical students increasingly recognize the significance
American healthcare system is not ranked among the best in quality of AI in their future daily practices, they understand the need for a more
[118] despite being the highest in GDP expenditure on healthcare [119]. profound knowledge of AI and its practical applications. Consequently,
Future studies could assess the impact of AI on the American healthcare future research should propose effective methods for integrating and
system due to the presence of startups. implementing AI within healthcare settings, adhering to a pedagogical
We also observe minimal investment in AI-based innovation startups process that fosters learning and skill development. This research must
in emerging economies and low-medium-income countries. Conse consider that medical curricula have long been known for their exten
quently, we recommend that future research concentrate on pinpointing sive and demanding nature [127,128], which may pose challenges in
12
incorporating AI-related content without exacerbating the existing Credit author statement
workload. To address these concerns, researchers should investigate
innovative teaching methodologies and curriculum design strategies AZ: Conceptualization, Methodology, Validation, Formal Analysis,
that facilitate the seamless integration of AI education into medical Investigation, Data Curation, Writing – Original Draft, Writing – Review
programs. These educational initiatives should equip medical students & Editing, Visualization, Project Administration, RPR: Validation,
with the knowledge and skills to harness AI’s potential effectively and Methodology, Supervision, Writing - Original Draft, Writing - Review &
ethically, ultimately enhancing patient care and outcomes within the Editing, DH: Validation, Methodology, Supervision, Writing - Original
rapidly evolving healthcare landscape. Draft, Writing - Review & Editing.
Data availability
6.3. Limitations
Data will be made available on request.
The current study has several limitations that should be acknowl
edged. First, by focusing only on Q1 journal publications, we may have References
overlooked valuable insights from articles published elsewhere. Second,
we relied on two primary databases, Scopus and PubMed; while these [1] L.G. Pee, S.L. Pan, L. Cui, Artificial intelligence in healthcare robots: a social
databases are reputable and comprehensive, other databases could have informatics study of knowledge embodiment, J. Assoc. Inf. Sci. Technol. 70 (4)
(2019) 351–369, https://doi.org/10.1002/asi.24145.
provided additional valuable information. Another limitation of our [2] S. Sunarti, F. Fadzlul Rahman, M. Naufal, M. Risky, K. Febriyanto, R. Masnina,
study stems from the strict criteria of selecting papers that directly Artificial intelligence in healthcare: opportunities and risk for future, Gac. Sanit.
answered our research questions. While this approach allowed for a 35 (2021) S67–S70, https://doi.org/10.1016/j.gaceta.2020.12.019.
[3] M. du Plessis, The role of knowledge management in innovation, J. Knowl.
focused analysis, it disregarded other papers that could have offered
Manag. 11 (4) (2007) 20–29, https://doi.org/10.1108/13673270710762684.
alternative insights and information. [4] A.M. Turing, Computing Machinery and Intelligence,” mind 59 (236) (1950)
Additionally, limiting our search to English-language publications 433–460, https://doi.org/10.1093/mind/LIX.236.433.
[5] B. Wahl, A. Cossy-Gantner, S. Germann, N.R. Schwalbe, Artificial intelligence (AI)
could have resulted in a less diverse representation of perspectives on AI
and global health: how can AI contribute to health in resource-poor settings? BMJ
innovation in healthcare. Lastly, only including peer-reviewed journal Glob. Health 3 (4) (2018), e000798 https://doi.org/10.1136/bmjgh-2018-
articles may have missed valuable information found in conference 000798.
proceedings, professional reports, or book chapters. Despite these limi [6] S. Hajkowicz, C. Sanderson, S. Karimi, A. Bratanova, C. Naughtin, “Artificial
Intelligence Adoption in the Physical Sciences, Natural Sciences, Life Sciences,
tations, our scoping review provides a comprehensive overview of AI Social Sciences and the Arts and Humanities: A Bibliometric Analysis of Research
innovation in healthcare from a global health perspective. Future studies Publications from 1960-2021,”, Technol. Soc., 2023, 102260 https://doi.org/
could address these limitations by expanding search criteria to include 10.1016/j.techsoc.2023.102260.
[7] C.H. Hoffmann, Is AI intelligent? An assessment of artificial intelligence, 70 years
other quarters, databases, languages, and a broader range of publication after Turing, Technol. Soc. 68 (2022), 101893, https://doi.org/10.1016/j.
types. techsoc.2022.101893.
[8] A. Bohr, K. Memarzadeh, Chapter 2 - the rise of artificial intelligence in
healthcare applications, in: A. Bohr, K. Memarzadeh (Eds.), Artificial Intelligence
7. Conclusions in Healthcare, Academic Press, 2020, pp. 25–60.
[9] X. Yang, Y. Wang, R. Byrne, G. Schneider, S. Yang, “Concepts of artificial
This paper rigorously examines the existing literature on AI inno intelligence for computer-assisted drug discovery,”, Chem. Rev. 119 (18) (2019)
10520–10594.
vation in healthcare and conducts an exploratory study of 2747 AI-based [10] P.M. Sánchez Sánchez, et al., Intelligent User Profiling Based on Sensors and
medical startups. The significant contributions of this paper include a Location Data to Detect Intrusions on Mobile Devices, 2020. https://www.
snapshot of the state of current research in the domain and suggest new igi-global.com/gateway/chapter/251794.
[11] T. Davenport, R. Kalakota, The potential for artificial intelligence in healthcare,
research directions to investigate and bring a holistic understanding of Future Healthc. J. 6 (2) (2019) 94–98, https://doi.org/10.7861/futurehosp.6-2-
ecosystem factors for AI healthcare startups and drivers of AI innovation 94.
and adoption in the healthcare sector. [12] A. de Barcelos Silva, et al., Intelligent personal assistants: a systematic literature
review, Expert Syst. Appl. 147 (2020), 113193, https://doi.org/10.1016/j.
We can observe from Table 3 that 75 papers appeared in 56 distinct eswa.2020.113193.
journals, which shows that AI in healthcare research is not confined to [13] T.H. Yimer, C. Wen, X. Yu, C. Jiang, A Study of the Minimum Safe Distance
any specific publication. This wide distribution of research across between Human Driven and Driverless Cars Using Safe Distance Model,
ArXiv200607022 Phys., 2020 [Online]. Available: http://arxiv.org/abs/2
various journals emphasizes the diverse, transversal, and interdisci 006.07022. (Accessed 26 May 2021).
plinary nature of AI applications in healthcare. It demonstrates the [14] A. Hosny, C. Parmar, J. Quackenbush, L.H. Schwartz, H.J.W.L. Aerts, Artificial
growing interest in this emerging topic from diverse perspectives. The intelligence in radiology, Nat. Rev. Cancer 18 (8) (2018), https://doi.org/
10.1038/s41568-018-0016-5.
many publications foster a more comprehensive understanding of the
[15] X.-Y. Zhou, Y. Guo, M. Shen, G.-Z. Yang, Application of artificial intelligence in
advances made in this area, further highlighting its significance in surgery, Front. Med. 14 (4) (2020) 417–430, https://doi.org/10.1007/s11684-
addressing complex healthcare challenges. 020-0770-0.
This paper acknowledges limitations associated with the quality of [16] J. French, et al., “Identification of patient prescribing predicting cancer diagnosis
using boosted decision trees,” in artificial intelligence in medicine, in: D. Riaño,
journal selection, criteria for the chosen publications, and language S. Wilk, A. ten Teije (Eds.), Lecture Notes in Computer Science, Springer
barriers. Moreover, we propose future research to address the paucity of International Publishing, Cham, 2019, pp. 328–333, https://doi.org/10.1007/
studies on AI healthcare startups, despite their rapid growth in the real 978-3-030-21642-9_42, 11526.
[17] M. Fakhoury, Artificial intelligence in psychiatry, in: Y.-K. Kim (Ed.), Frontiers in
world. We advocate for increased collaboration to improve research Psychiatry: Artificial Intelligence, Precision Medicine, and Other Paradigm Shifts,
efforts in low- and middle-income countries (LMICs) and within man Advances in Experimental Medicine and Biology, Springer, Singapore, 2019,
agement studies related to AI healthcare innovation. Additionally, we pp. 119–125, https://doi.org/10.1007/978-981-32-9721-0_6.
[18] M.R. Neves, D.W.R. Marsh, Modelling the impact of AI for clinical decision
recommend enhancing research efforts to identify innovation factors support, in: D. Riaño, S. Wilk, A. ten Teije (Eds.), Artificial Intelligence in
and develop adoption models specific to AI healthcare, further enriching Medicine, Lecture Notes in Computer Science, vol. 11526, Springer International
our understanding and application of these transformative technologies. Publishing, Cham, 2019, pp. 292–297, https://doi.org/10.1007/978-3-030-
21642-9_37.
[19] R. Bhatia, Telehealth and COVID-19: using technology to accelerate the curve on
Funding sources access and quality healthcare for citizens in India, Technol. Soc. 64 (2021),
101465, https://doi.org/10.1016/j.techsoc.2020.101465.
[20] A. Kasthuri, Challenges to healthcare in India - the five A’s, Indian J. Community
This research received no specific grant from funding agencies in the Med. Off. Publ. Indian Assoc. Prev. Soc. Med. 43 (3) (2018) 141–143, https://doi.
public, commercial, or not-for-profit sectors. org/10.4103/ijcm.IJCM_194_18.
13
[21] I. of M. (US) C. on the H. P. E. Summit, A. C. Greiner, E. Knebel, Challenges Facing [44] E. Chikhaoui, A. Alajmi, S. Larabi-Marie-Sainte, Artificial intelligence
the Health System and Implications for Educational Reform, National Academies applications in healthcare sector: ethical and legal challenges, Emerg. Sci. J. 6 (4)
Press (US), 2003. https://www.ncbi.nlm.nih.gov/books/NBK221522/. (2022), https://doi.org/10.28991/ESJ-2022-06-04-05.
[22] R. Zhidkoff, “the future of healthcare with AI,” medium, 23, https://becominghu [45] K. Paranjape, M. Schinkel, R. Nannan Panday, J. Car, P. Nanayakkara,
man.ai/the-future-of-healthcare-with-ai-f8593f877b08, 2020. Introducing artificial intelligence training in medical education, JMIR Med. Educ.
[23] N.R. Sahni, G. Stein, R. Zemmel, D.M. Cutler, The potential impact of artificial 5 (2) (2019), e16048, https://doi.org/10.2196/16048.
intelligence on healthcare spending, in: The Economics of Artificial Intelligence: [46] L.G. McCoy, S. Nagaraj, F. Morgado, V. Harish, S. Das, L.A. Celi, What do medical
Health Care Challenges, University of Chicago Press, 2023. https://www.nber.or students actually need to know about artificial intelligence? Npj Digit. Med. 3 (1)
g/books-and-chapters/economics-artificial-intelligence-health-care-challenge (2020) 86, https://doi.org/10.1038/s41746-020-0294-7.
s/potential-impact-artificial-intelligence-healthcare-spending. [47] S.Y. Chai, A. Hayat, G.T. Flaherty, Integrating artificial intelligence into
[24] G. Rong, A. Mendez, E. Bou Assi, B. Zhao, M. Sawan, Artificial intelligence in haematology training and practice: opportunities, threats and proposed solutions,
healthcare: review and prediction case studies, Eng. Times 6 (3) (2020) 291–301, Br. J. Haematol. 198 (5) (2022) 807–811, https://doi.org/10.1111/bjh.18343.
https://doi.org/10.1016/j.eng.2019.08.015. [48] A. Boonstra, M. Laven, Influence of artificial intelligence on the work design of
[25] J. Radhakrishnan, M. Chattopadhyay, Determinants and barriers of artificial emergency department clinicians a systematic literature review, BMC Health
intelligence adoption – a literature review, in: S.K. Sharma, Y.K. Dwivedi, Serv. Res. 22 (1) (May 2022) 669, https://doi.org/10.1186/s12913-022-08070-7.
B. Metri, N.P. Rana (Eds.), Re-imagining Diffusion and Adoption of Information [49] B. Gong, et al., Influence of artificial intelligence on Canadian medical students’
Technology and Systems: A Continuing Conversation, IFIP Advances in preference for radiology specialty: ANational survey study, Acad. Radiol. 26 (4)
Information and Communication Technology, Springer International Publishing, (2019) 566–577, https://doi.org/10.1016/j.acra.2018.10.007.
Cham, 2020, pp. 89–99, https://doi.org/10.1007/978-3-030-64849-7_9. [50] C. Sit, et al., Attitudes and perceptions of UK medical students towards artificial
[26] M.G. Colombo, E. Piva, Strengths and weaknesses of academic startups: a intelligence and radiology: a multicentre survey, Insights Imaging 11 (1) (2020)
conceptual model, IEEE Trans. Eng. Manag. 55 (1) (2008) 37–49, https://doi.org/ 14, https://doi.org/10.1186/s13244-019-0830-7.
10.1109/TEM.2007.912807. [51] I. Bartoletti, AI in healthcare: ethical and privacy challenges, in: D. Riaño, S. Wilk,
[27] A. Davila, G. Foster, M. Gupta, Venture capital financing and the growth of A. ten Teije (Eds.), Artificial Intelligence in Medicine, Lecture Notes in Computer
startup firms, J. Bus. Ventur. 18 (6) (2003) 689–708, https://doi.org/10.1016/ Science, vol. 11526, Springer International Publishing, Cham, 2019, pp. 7–10,
S0883-9026(02)00127-1. https://doi.org/10.1007/978-3-030-21642-9_2.
[28] A. Kuckertz, J. Block, Reviewing systematic literature reviews: ten key questions [52] O. Asan, A.E. Bayrak, A. Choudhury, Artificial intelligence and human trust in
and criteria for reviewers, Manag. Rev. Q. 71 (3) (2021) 519–524, https://doi. healthcare: focus on clinicians (preprint), J. Med. Internet Res. (2019), https://
org/10.1007/s11301-021-00228-7. doi.org/10.2196/preprints.15154 preprint.
[29] D. Moher, A. Liberati, J. Tetzlaff, D.G. Altman, T.P. Group, Preferred reporting [53] E. Vayena, A. Blasimme, I.G. Cohen, Machine learning in medicine: addressing
Items for systematic reviews and meta-analyses: the PRISMA statement, PLoS ethical challenges, PLoS Med. 15 (11) (2018), e1002689, https://doi.org/
Med. 6 (7) (2009), e1000097, https://doi.org/10.1371/journal.pmed.1000097. 10.1371/journal.pmed.1002689.
[30] P. Ossom Williamson, C.I.J. Minter, Exploring PubMed as a reliable resource for [54] D. Misra, et al., Early detection of septic shock onset using interpretable machine
scholarly communications services, J. Med. Libr. Assoc. JMLA 107 (1) (2019) learners, J. Clin. Med. 10 (2) (2021), https://doi.org/10.3390/jcm10020301.
16–29, https://doi.org/10.5195/jmla.2019.433. [55] J. Feng, et al., Clinical artificial intelligence quality improvement: towards
[31] M.E. Falagas, E.I. Pitsouni, G.A. Malietzis, G. Pappas, Comparison of PubMed, continual monitoring and updating of AI algorithms in healthcare, Npj Digit.
Scopus, Web of science, and Google scholar: strengths and weaknesses, The Med. 5 (1) (2022), https://doi.org/10.1038/s41746-022-00611-y.
FASEB Journal - Wiley Online Library 22 (2) (2008), https://doi.org/10.1096/ [56] S. Nundy, T. Montgomery, R.M. Wachter, Promoting trust between patients and
fj.07-9492LSF. physicians in the era of artificial intelligence, JAMA 322 (6) (2019) 497, https://
[32] D. Pinto dos Santos, et al., Medical students’ attitude towards artificial doi.org/10.1001/jama.2018.20563.
intelligence: a multicentre survey, Eur. Radiol. 29 (4) (2019) 1640–1646, https:// [57] C. Longoni, C.K. Morewedge, AI can outperform doctors. So why don’t patients
doi.org/10.1007/s00330-018-5601-1. trust it? Harv. Bus. Rev. (2019). https://hbr.org/2019/10/ai-can-outperform-
[33] N. Coudray, et al., Classification and mutation prediction from non–small cell doctors-so-why-dont-patients-trust-it.
lung cancer histopathology images using deep learning, Nat. Med. 24 (10) (2018), [58] F. Gille, A. Jobin, M. Ienca, What we talk about when we talk about trust: theory
https://doi.org/10.1038/s41591-018-0177-5. of trust for AI in healthcare, Intell.-Based Med. 1 (2) (2020), 100001, https://doi.
[34] F. Collado-Mesa, E. Alvarez, K. Arheart, The role of artificial intelligence in org/10.1016/j.ibmed.2020.100001.
diagnostic radiology: a survey at a single radiology residency training program, [59] F. Jiang, et al., Artificial intelligence in healthcare: past, present and future,
J. Am. Coll. Radiol. 15 (12) (2018) 1753–1757, https://doi.org/10.1016/j. Stroke Vasc. Neurol. 2 (4) (2017) 230–243, https://doi.org/10.1136/svn-2017-
jacr.2017.12.021. 000101.
[35] J.N. Kather, et al., Deep learning can predict microsatellite instability directly [60] M.-T. Ho, N.-T.B. Le, P. Mantello, M.-T. Ho, N. Ghotbi, Understanding the
from histology in gastrointestinal cancer, Nat. Med. 25 (7) (2019), https://doi. acceptance of emotional artificial intelligence in Japanese healthcare system: a
org/10.1038/s41591-019-0462-y. cross-sectional survey of clinic visitors’ attitude, Technol. Soc. 72 (2023),
[36] H. Alami, et al., Artificial intelligence in health care: laying the Foundation for 102166, https://doi.org/10.1016/j.techsoc.2022.102166.
Responsible, sustainable, and inclusive innovation in low- and middle-income [61] D.S. Liu, K. Abu-Shaban, S.S. Halabi, T.S. Cook, Changes in radiology due to
countries, Glob. Health 16 (1) (2020) 52, https://doi.org/10.1186/s12992-020- artificial intelligence that can attract medical students to the specialty, JMIR Med.
00584-1. Educ. 9 (1) (2023), e43415, https://doi.org/10.2196/43415.
[37] J.-A. Moldt, T. Festl-Wietek, A. Madany Mamlouk, K. Nieselt, W. Fuhl, [62] K.-H. Yu, A.L. Beam, I.S. Kohane, Artificial intelligence in healthcare, Nat.
A. Herrmann-Werner, Chatbots for future docs: exploring medical students’ Biomed. Eng. 2 (10) (Oct. 2018) 719–731, https://doi.org/10.1038/s41551-018-
attitudes and knowledge towards artificial intelligence and medical chatbots, 0305-z.
Med. Educ. Online 28 (1) (2023), 2182659, https://doi.org/10.1080/ [63] F. Luna-Perejon, et al., An automated fall detection system using recurrent neural
10872981.2023.2182659. networks, in: D. Riaño, S. Wilk, A. ten Teije (Eds.), Artificial Intelligence in
[38] M. Garbuio, N. Lin, Artificial intelligence as a growth engine for health care Medicine, Lecture Notes in Computer Science, vol. 11526, Springer International
startups: emerging business models, Calif. Manage. Rev. 61 (2) (2019) 59–83, Publishing, Cham, 2019, pp. 36–41, https://doi.org/10.1007/978-3-030-21642-
https://doi.org/10.1177/0008125618811931. 9_6.
[39] B. Ehteshami Bejnordi, et al., Diagnostic assessment of deep learning algorithms [64] C.C. Ghosh, D. McVicar, G. Davidson, C. Shannon, C. Armour, What can we learn
for detection of lymph node metastases in w omen with breast cancer, JAMA 318 about the psychiatric diagnostic categories by analysing patients’ lived
(22) (2017), https://doi.org/10.1001/jama.2017.14585. experiences with Machine-Learning? BMC Psychiatr. 22 (1) (2022) 427, https://
[40] X. Du-Harpur, F.M. Watt, N.M. Luscombe, M.D. Lynch, What is AI? Applications doi.org/10.1186/s12888-022-03984-2.
of artificial intelligence to dermatology, Br. J. Dermatol. 183 (3) (2020) 423–430, [65] L. Tzelves, et al., Artificial intelligence supporting cancer patients across
https://doi.org/10.1111/bjd.18880. Europe—the ASCAPE project, PLoS One 17 (4) (2022), e0265127, https://doi.
[41] C. Hettiarachchi, C. Chitraranjan, A machine learning approach to predict org/10.1371/journal.pone.0265127.
diabetes using short recorded photoplethysmography and physiological [66] D. Thesmar, D. Sraer, L. Pinheiro, N. Dadson, R. Veliche, P. Greenberg,
characteristics, in: D. Riaño, S. Wilk, A. ten Teije (Eds.), Artificial Intelligence in Combining the power of artificial intelligence with the richness of healthcare
Medicine, Lecture Notes in Computer Science, vol. 11526, Springer International claims data: opportunities and challenges, Pharmacoeconomics 37 (6) (2019)
Publishing, Cham, 2019, pp. 322–327, https://doi.org/10.1007/978-3-030- 745–752, https://doi.org/10.1007/s40273-019-00777-6.
21642-9_41. [67] S.A. Wartman, C.D. Combs, Medical education must move from the information
[42] R. Ben-Ari, Y. Shoshan, T. Tlusty, Mammogram classification with ordered loss, age to the age of artificial intelligence, Acad. Med. 93 (8) (2018) 1107–1109,
in: D. Riaño, S. Wilk, A. ten Teije (Eds.), Artificial Intelligence in Medicine, https://doi.org/10.1097/ACM.0000000000002044.
Lecture Notes in Computer Science, vol. 11526, Springer International [68] S. Secinaro, D. Calandra, A. Secinaro, V. Muthurangu, P. Biancone, The role of
Publishing, Cham, 2019, pp. 67–76, https://doi.org/10.1007/978-3-030-21642- artificial intelligence in healthcare: a structured literature review, BMC Med. Inf.
9_10. Decis. Making 21 (1) (2021) 125, https://doi.org/10.1186/s12911-021-01488-9.
[43] V. Menger, M. Spruit, W. van der Klift, F. Scheepers, Using cluster ensembles to [69] N.E. Sharpless, A.R. Kerlavage, The potential of AI in cancer care and research,
identify psychiatric patient subgroups, in: D. Riaño, S. Wilk, A. ten Teije (Eds.), Biochim. Biophys. Acta BBA - Rev. Cancer 1876 (2021), 188573, https://doi.org/
Artificial Intelligence in Medicine, Lecture Notes in Computer Science, vol. 10.1016/j.bbcan.2021.188573.
11526, Springer International Publishing, Cham, 2019, pp. 252–262, https://doi. [70] J.M. Tenório, A.D. Hummel, F.M. Cohrs, V.L. Sdepanian, I.T. Pisa, H. de Fátima
org/10.1007/978-3-030-21642-9_31. Marin, Artificial intelligence techniques applied to the development of a decision-
14
support system for diagnosing celiac disease, Int. J. Med. Inf. 80 (11) (2011) [98] A.S. Young, AI in healthcare startups and special challenges, Intell.-Based Med. 6
793–802, https://doi.org/10.1016/j.ijmedinf.2011.08.001. (Jan. 2022), 100050, https://doi.org/10.1016/j.ibmed.2022.100050.
[71] S.P. Somashekhar, et al., Watson for Oncology and breast cancer treatment [99] “With mobile monitoring for cardiac patients, help is just a heartbeat away,”
recommendations: agreement with an expert multidisciplinary tumor board, Ann. Transform. https://news.microsoft.com/transform/videos/with-mobile-moni
Oncol. Off. J. Eur. Soc. Med. Oncol. 29 (2) (2018) 418–423, https://doi.org/ toring-for-cardiac-patients-help-is-just-a-heartbeat-away/.
10.1093/annonc/mdx781. [100] J.-P. Tsang, S. Mohan, Benefits of monitoring patients with mobile cardiac
[72] D. Mincu, S. Roy, Developing robust benchmarks for driving forward AI telemetry (MCT) compared with the Event or Holter monitors, Med. Devices
innovation in healthcare, Nat. Mach. Intell. 4 (11) (2022), https://doi.org/ Auckl. NZ 7 (2013) 1–5, https://doi.org/10.2147/MDER.S54038.
10.1038/s42256-022-00559-4. [101] K.D. Starič, V. Trajkovik, H. Belani, A. Vitagliano, P. Bukovec, Smart phone
[73] A. Shademan, R.S. Decker, J.D. Opfermann, S. Leonard, A. Krieger, P.C.W. Kim, applications for self-monitoring of the menstrual cycle: a review and content
Supervised autonomous robotic soft tissue surgery, Sci. Transl. Med. 8 (337) analysis, Clin. Exp. Obstet. Gynecol. 46 (5) (2019), https://doi.org/10.12891/
(2016), https://doi.org/10.1126/scitranslmed.aad9398, 337ra64-337ra64. ceog4830.2019.
[74] G. Bunney, et al., Using machine learning to predict hospital disposition with [102] A. Lal, J. Dang, C. Nabzdyk, O. Gajic, V. Herasevich, Regulatory oversight and
geriatric emergency department innovation intervention, Ann. Emerg. Med. 81 ethical concerns surrounding software as medical device (SaMD) and digital twin
(3) (2023) 353–363, https://doi.org/10.1016/j.annemergmed.2022.07.026. technology in healthcare, Ann. Transl. Med. 10 (18) (2022), https://doi.org/
[75] M. Chen, et al., Acceptance of clinical artificial intelligence among physicians and 10.21037/atm-22-4203.
medical students: a systematic review with cross-sectional survey, Front. Med. 9 [103] J.D. Lee, K. A., See, “trust in automation: designing for appropriate reliance,”,
(2022). https://www.frontiersin.org/articles/10.3389/fmed.2022.990604. Hum. Factors 46 (1) (2004) 50–80, https://doi.org/10.1518/hfes.46.1.50_30392.
[76] M.D. Abràmoff, et al., A reimbursement framework for artificial intelligence in [104] M. Promberger, J. Baron, Do patients trust computers? J. Behav. Decis. Making 19
healthcare, Npj Digit. Med. 5 (1) (2022), https://doi.org/10.1038/s41746-022- (5) (2006) 455–468, https://doi.org/10.1002/bdm.542.
00621-w. [105] M. Fenech, N. Strukelj, O. Buston, “Ethical, Social, and Political Challenges of
[77] J.P. Pestian, et al., A machine learning approach to identifying the thought Artificial Intelligence in Health,”, Future Advocacy, https://futureadvocacy.co
markers of suicidal subjects: a prospective multicenter trial, Suicide Life- m/publications/ethical-social-and-political-challenges-of-artificial-intelligen
Threatening Behav. 47 (1) (2017) 112–121, https://doi.org/10.1111/sltb.12312. ce-in-health/.
[78] M.W. Mulholland, P. Abrahamse, V. Bahl, Linear programming to optimize [106] J. Yuan, et al., The tip of the ‘celiac iceberg’ in China: a systematic review and
performance in a department of surgery, J. Am. Coll. Surg. 200 (6) (2005) meta-analysis, PLoS One 8 (12) (Dec. 2013), e81151, https://doi.org/10.1371/
861–868, https://doi.org/10.1016/j.jamcollsurg.2005.01.001. journal.pone.0081151.
[79] A.A. Miyazawa, Artificial intelligence: the future for cardiology, Heart 105 (15) [107] M. Hackett, Trustworthy AI has the ability to transform healthcare,
(2019), https://doi.org/10.1136/heartjnl-2018-314464, 1214–1214. MobiHealthNews, Jan. 12 (2021). https://www.mobihealthnews.com/news/tr
[80] F. Piccialli, F. Giampaolo, E. Prezioso, D. Camacho, G. Acampora, Artificial ustworthy-ai-has-ability-transform-healthcare.
intelligence and healthcare: forecasting of medical bookings through multi-source [108] C. for D. and R. Health, Artificial Intelligence and Machine Learning in Software
time-series fusion, Inf. Fusion 74 (2021) 1–16, https://doi.org/10.1016/j. as a Medical Device, FDA, Jan. 2021. https://www.fda.gov/medical-devices/soft
inffus.2021.03.004. ware-medical-device-samd/artificial-intelligence-and-machine-learning-software
[81] J.-H. Lee, D.-H. Kim, S.-N. Jeong, S.-H. Choi, Detection and diagnosis of dental -medical-device.
caries using a deep learning-based convolutional neural network algorithm, [109] C.M. Pfeifer, A progressive three-phase innovation to medical education in the
J. Dent. 77 (2018) 106–111, https://doi.org/10.1016/j.jdent.2018.07.015. United States, Med. Educ. Online 23 (1) (2018), 1427988, https://doi.org/
[82] K.R. Laukamp, et al., Fully automated detection and segmentation of 10.1080/10872981.2018.1427988.
meningiomas using deep learning on routine multiparametric MRI, Eur. Radiol. [110] N. Imran, M. Jawaid, Artificial intelligence in medical education: are we ready for
29 (1) (2019) 124–132, https://doi.org/10.1007/s00330-018-5595-8. it? Pakistan J. Med. Sci. 36 (5) (2020) https://doi.org/10.12669/pjms.36.5.3042.
[83] V.B. Kolachalama, Machine learning and pre-medical education, Artif. Intell. [111] K. Masters, Artificial Intelligence developments in medical education: a
Med. 129 (2022), 102313, https://doi.org/10.1016/j.artmed.2022.102313. conceptual and practical framework, MedEdPublish 9 (2020), https://doi.org/
[84] A. Esteva, et al., Dermatologist-level classification of skin cancer with deep neural 10.15694/mep.2020.000239.1.
networks, Nature 542 (7639) (2017) 115–118, https://doi.org/10.1038/ [112] T.P. Caudell, et al., Virtual patient simulator for distributed collaborative medical
nature21056. education, Anat. Rec. B. New Anat. 270B (1) (2003) 23–29, https://doi.org/
[85] S. McLennan, A. Fiske, D. Tigard, R. Müller, S. Haddadin, A. Buyx, Embedded 10.1002/ar.b.10007.
ethics: a proposal for integrating ethics into the development of medical AI, BMC [113] “Home,” Carle Illinois College of Medicine..https://medicine.illinois.ed
Med. Ethics 23 (1) (2022) 6, https://doi.org/10.1186/s12910-022-00746-3. u/(accessed May 07, 2021).
[86] P. Kumar, S.K. Sharma, V. Dutot, Artificial intelligence (AI)-enabled CRM [114] “Center for Artificial Intelligence in Medicine & Imaging |.”.https://aimi.stanford.
capability in healthcare: the impact on service innovation, Int. J. Inf. Manag. 69 edu/(accessed May 07, 2021).
(2023), 102598, https://doi.org/10.1016/j.ijinfomgt.2022.102598. [115] “Crunchbase: Discover innovative companies and the people behind them.”.
[87] E. Choi, A. Schuetz, W.F. Stewart, J. Sun, Using recurrent neural network models https://www.crunchbase.com/(accessed Jun. 20, 2023).
for early detection of heart failure onset, J. Am. Med. Inform. Assoc. JAMIA 24 (2) [116] “CB Insights - Technology Market Intelligence,” CB Insights..https://www.cbinsi
(2017) 361–370, https://doi.org/10.1093/jamia/ocw112. ghts.com/(accessed May 24, 2023).
[88] G. Bedi, et al., Automated analysis of free speech predicts psychosis onset in high- [117] V. Galaz, et al., Artificial intelligence, systemic risks, and sustainability, Technol.
risk youths, Npj Schizophr 1 (1) (2015), 15030, https://doi.org/10.1038/ Soc. 67 (2021), 101741, https://doi.org/10.1016/j.techsoc.2021.101741.
npjschz.2015.30. [118] “Best Healthcare in the World 2023,” Wisevoter. https://wisevoter.com/country
[89] Z.I. Attia, D.M. Harmon, E.R. Behr, P.A. Friedman, Application of artificial -rankings/best-healthcare-in-the-world/(accessed Jun. 10, 2023).
intelligence to the electrocardiogram, Eur. Heart J. 42 (46) (2021) 4717–4730, [119] I. Papanicolas, L.R. Woskie, A.K. Jha, Health care spending in the United States
https://doi.org/10.1093/eurheartj/ehab649. and other high-income countries, JAMA 319 (10) (Mar. 2018) 1024–1039,
[90] B. Meskó, G. Hetényi, Z. Győrffy, Will artificial intelligence solve the human https://doi.org/10.1001/jama.2018.1150.
resource crisis in healthcare? BMC Health Serv. Res. 18 (1) (2018) 545, https:// [120] T. Akenroye, Factors influencing innovation in healthcare: a conceptual synthesis,
doi.org/10.1186/s12913-018-3359-4. Innov. J. 17 (Jan. 2012).
[91] P. Kuo, R. Schroeder, S. Mahaffey, R. Bollinger, Optimization of operating room [121] F.K. Al Badi, K.A. Alhosani, F. Jabeen, A. Stachowicz-Stanusch, N. Shehzad,
allocation using linear programming techniques, J. Am. Coll. Surg. (2003), W. Amann, Challenges of AI adoption in the UAE healthcare, Vis. J. Bus. Perspect.
https://doi.org/10.1016/J.JAMCOLLSURG.2003.07.006. (2021), 097226292098839, https://doi.org/10.1177/0972262920988398.
[92] Y.M. Chiu, J. Courteau, I. Dufour, A. Vanasse, C. Hudon, Machine learning to [122] H. Haider, “Barriers to the adoption of artificial intelligence in Healthcare in
improve frequent emergency department use prediction: a retrospective cohort India,” OpenDocs Home, https://opendocs.ids.ac.uk/opendocs/handle
study, Sci. Rep. 13 (1) (2023), https://doi.org/10.1038/s41598-023-27568-6. /20.500.12413/15272.
[93] G. Dicuonzo, F. Donofrio, A. Fusco, M. Shini, Healthcare System: Moving Forward [123] F.D. Davis, Perceived usefulness, perceived ease of use, and user acceptance of
with Artificial Intelligence, 2022, 102510, https://doi.org/10.1016/j. information technology, MIS Q. 13 (3) (1989) 319–340, https://doi.org/10.2307/
technovation.2022.102510. Technovation. 249008.
[94] C. Cannavale, A. Esempio Tammaro, D. Leone, F. Schiavone, Innovation adoption [124] V. Venkatesh, M.G. Morris, G.B. Davis, F.D. Davis, User acceptance of information
in inter-organizational healthcare networks – the role of artificial intelligence, technology: toward a unified view, MIS Q. 27 (3) (2003) 425–478, https://doi.
Eur. J. Innovat. Manag. 25 (6) (2022) 758–774, https://doi.org/10.1108/EJIM- org/10.2307/30036540.
08-2021-0378. [125] M.U.H. Uzir, et al., Applied Artificial Intelligence and user satisfaction:
[95] I. Kulkov, Next-generation business models for artificial intelligence startups in smartwatch usage for healthcare in Bangladesh during COVID-19, Technol. Soc.
the healthcare industry, Int. J. Entrepreneurial Behav. Res. (2021), https://doi. 67 (2021), 101780, https://doi.org/10.1016/j.techsoc.2021.101780.
org/10.1108/IJEBR-04-2021-0304. [126] X. Liu, X. He, M. Wang, H. Shen, What influences patients’ continuance intention
[96] L. Petersson, et al., Challenges to implementing artificial intelligence in to use AI-powered service robots at hospitals? The role of individual
healthcare: a qualitative interview study with healthcare leaders in Sweden, BMC characteristics, Technol. Soc. 70 (2022), 101996, https://doi.org/10.1016/j.
Health Serv. Res. 22 (1) (2022) 850, https://doi.org/10.1186/s12913-022- techsoc.2022.101996.
08215-8. [127] G. Bordage, The curriculum: overloaded and too general? Med. Educ. 21 (3)
[97] H. Alami, et al., Artificial intelligence and health technology assessment: (1987) 183–188, https://doi.org/10.1111/j.1365-2923.1987.tb00689.x.
anticipating a new level of complexity, J. Med. Internet Res. 22 (7) (Jul. 2020), [128] S. Slavin, M.F. D’Eon, Overcrowded curriculum is an impediment to change (Part
e17707, https://doi.org/10.2196/17707. A), Can. Med. Educ. J. 12 (4) (2021), https://doi.org/10.36834/cmej.73532.
15

1 s2.0 S0160791X23001264 Main

Uploaded by

Copyright:

Available Formats

1 s2.0 S0160791X23001264 Main

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 S0160791X23001264 Main

Uploaded by

Copyright:

Available Formats

Technology in Society 74 (2023) 102321

Contents lists available at ScienceDirect

Artificial intelligence innovation in healthcare: Literature review,

1. Introduction reliable e-data to drive algorithm development [9]. In healthcare, data

Fig. 2. Literature search and identification procedures.

Fig. 3. “Connectedpapers.com” tool for papers unrelated to the research [38].

Fig. 5. Number of publications per year in our literature.

Table 3 Table 3 (continued )

Table 6 years of medical school, general training, internships, and a time of

ideas [45]. diagnostics and early detection. As an overview of these startups, we

Fig. 6. Number of AI healthcare startups founded per year.

You might also like