According to the Regulations of the People’s Republic of China on Ethical Review of Science and Technology (Trial), Number 167 of the State Science and Technology Development Supervision (2023), scientific research activities involving humans or other animals need to undergo ethical review. This thesis does not involve humans or other animals, nor does it pose risks to life and health, the ecological environment, public order, or sustainable development. Therefore, ethical approval is not required.

Figure 1 shows that a total of 195 papers on AI and medical education have been published in the past decade, showing an overall upward trend. The publications saw a significant surge from 2020 to 2021, reaching a peak in 2021, although the number of related papers published in 2022 decreased. The development of AI presented unprecedented opportunities and challenges to the medical and health industry. Medical education, being the cornerstone of medical industry development, can benefit from the application of AI, driving continual innovation.

Based on a comprehensive national analysis, 57 countries globally contributed to the exploration of AI within the field of medical education from 2013 to 2022. The United States took the lead by publishing 66 papers, thereby establishing itself as the most actively engaged country in this domain. The subsequent countries, albeit with lesser contributions, were Canada (24 papers), China (17 papers), England (13 papers), Singapore (12 papers), Australia (12 papers), India (9 papers), Germany (8 papers), the Netherlands (8 papers), and Spain (7 papers). The most cited countries were the United States (845 citations), Singapore (489 citations), and China (435 citations). When evaluated in terms of total link strength, the United States (44), the Netherlands (29), and Belgium (26) emerged as the top 3 countries (Table 2). Figure 2 shows that a clear inclination of North American and European countries toward the application of AI in medical education is evident, possibly due to their technological advancement. The United States has been a front-runner in this arena, publishing a multitude of relevant papers. Concurrently, it has fostered collaborative relationships with various countries for related research.

Shifting the focus to an institutional analysis reveals that from 2013 to 2022, 77 institutions were engaged in research on AI in medical education. The two institutions that topped the list in terms of the number of publications were Harvard Medical School and the University of Toronto, each with 7 contributions, followed by McGill University and the National University of California, San Francisco (5 contributions each) (Table 3). The institutions receiving the most citations were Nanyang Technological University (396 citations), McGill University (149 citations), and the University of Chicago (127 citations). Figure 3 shows that Leiden University and Harvard Medical School demonstrated more collaboration with other institutions, both exhibiting a link strength of 15.

In the span of the last decade, research on AI and medical education has seen the involvement of a total of 53 authors. The authors most frequently contributing to the documents included Vincent Bissonnette, Charlotte Blacketer, Rolando F Del Maestro, Nicole Ledwos, Nykan Mirchi, Alexander Winkler-Schwartz, and Recai Yilmaz, each writing 3 papers. The authors garnering the highest citations encompassed the same group, with each achieving 143 citations (Table 4). As discerned from the VOSviewer image, there are no researchers with a significantly high number of publications, indicating that the volume of published papers remains relatively minimal. Figure 4 shows that research in this field is still nascent, with no particular research team outperforming others.

In accordance with Table 5, there are 15 papers that serve as primary references in the research of AI and medical education. The paper titled “Medical Students’ Attitude Towards Artificial Intelligence: A Multicenter Survey” emerged as the most frequently cited and most pertinent literature, garnering 36 and 109 citations, respectively. It primarily evaluates the attitudes of undergraduate medical students toward radiology and medical AI.

The papers “Machine Learning and Medical Education” and “Attitudes and Perceptions of UK Medical Students Towards Artificial Intelligence and Radiology: A Multicenter Survey” are the second most frequently cited. The papers “Medical Education Must Move From the Information Age to the Age of Artificial Intelligence” and “Machine Learning and Medical Education” occupy the second position in terms of total link strength. Figure 5 illustrates this information.

The study examining AI and medical education from 2013 to 2022 concentrated on 39 primary keywords (Table 6). Figure 6 shows that AI (100), education (47), and medical education (45) have the highest frequency and connection intensity.

Figure 7 shows that the analysis of references with high citation frequency and centrality enables us to understand highly respected research results in the application of AI in medical education.

In clusters 0 and 1, the swift advancement of AI has led to its application across all medical sectors, notably radiology [25-27]. Despite radiologists, residents, and medical students increasingly recognizing the importance of understanding AI, medical education that targets future radiologists is only just commencing [14,20,28]. Current investigations fall into 3 categories, that are (1) methods to facilitate medical students in learning AI knowledge, (2) using AI technology to augment radiology teaching efficiency and assist medical students in identifying clinical images, and (3) medical students’ attitudes toward AI application in radiology. An AI curriculum (Artificial Intelligence in Radiology [AI-RADS]) has been devised to equip residents devoid of computing backgrounds with basic AI knowledge and its radiology application. The curriculum was highly rated (9.8 out of 10) by residents for overall satisfaction and significantly increased students’ confidence in interpreting AI-related journal papers. There was a marked improvement in residents’ comprehension of AI’s fundamental concepts [29]. Some institutions emphasize integrating AI frameworks to strengthen radiology education. For example, after scanning, the patient's condition will be interpreted by artificial intelligence to give a preliminary diagnosis. AI assigns cases to interns whose personal profiles indicate that they will benefit the most. Interns cooperate with artificial intelligence and use equivalent tools for diagnosis. Interns and attending radiologists elaborate on the final report. AI uses natural language processing to anonymize new cases, add them to the teaching archive, and update the personal profiles of trainees after new cases are completed. When trainees review cases similar to new cases, AI will provide them with corresponding cases from the teaching archive.[30]. As this framework continues to evolve, it may be possible to achieve “precise medical education” tailored to the individual learning styles and needs of the students [30]. A multicenter survey assessing UK medical students’ attitudes and perceptions of AI and radiology revealed that students recognize the significance of AI and are eager to engage [17]. This prompts the need to integrate relevant AI courses into medical education to acquaint students with practical AI applications and constraints, thereby maintaining their learning enthusiasm and preventing AI-related panic.

Natural language processing is an important direction in the fields of computer science and AI. It studies various theories and methods that enable effective communication between humans and computers using natural language. Its main function here is to distinguish rare cases

In cluster 2, eHealth refers to the use of information and communication technologies to fulfill health care needs in various domains, including AI, telemedicine, Internet of Things, connected devices, and mobile health (mHealth) [31]. eHealth technologies provide access to health care in rural areas and support the management of numerous health conditions [32-36]. Following the release of the World Health Organization’s national eHealth strategy tool in 2012, it is imperative for future medical students to receive eHealth education and training. Current medical education primarily includes conceptual courses while neglecting practical training [37]. While emphasizing the inclusion of eHealth in medical education, it is also important to recognize the potential adverse outcomes of over-reliance on AI technology [38]. Hence, identifying the optimal eHealth application areas in health care is necessary [39].

In cluster 3, the integration of medical education and AI holds significant value and potential beyond radiology, extending into surgical education and surgery. AI’s earliest medical applications were in image-based specialties, such as radiology, pathology, ophthalmology, and dermatology. However, its application in procedural professions such as surgery may require more time [40,41]. The benefits of AI application in surgery mainly include integrating preoperative, intraoperative, and postoperative data to improve the accuracy of the clinical decision-making system and predict postoperative complications more efficiently and applying surgical knowledge and education to interact with surgeons and patients through virtual or augmented reality. For instance, virtual reality simulators were initially used in laparoscopic surgery training [42]. A study involving 176 medical students was conducted to assess the accuracy of robot-assisted virtual surgical simulations after integrated deep learning, showing improved accuracy [43]. In 2022 and 2023, AI application breakthroughs were achieved in oral and maxillofacial surgery education [44] and orthopedic surgery [45]. While AI proves beneficial in surgery and surgical education, especially in surgical ability assessment, it raises questions regarding whether AI can ever match the intelligence and audacity of the human educators. Although advanced AI teaching tools can be incorporated into surgical education, current technology cannot fully replace multifaceted surgeons or surgical educators. Addressing the transparency and responsibility of AI application in medical education and resolving ethical issues may require more time and effort.

In cluster 5, the rapid AI development profoundly impacts medical education. Modern medical education must accommodate various health care systems, including digital health systems and big data generation in a highly connected world [46]. A Canadian survey of medical students’ perceptions of AI’s impact on radiology in 2018 showed that anxiety induced by the prospect of AI replacing radiologists deterred many students from considering radiology [18]. The radiology community should appreciate AI’s potential impact on the profession, educate students appropriately about AI’s role, and ensure radiology’s viability as a long-term career option. While AI’s benefits in medicine include eliminating human bias and enhancing pattern recognition and decision-making, its drawbacks, such as the inability to provide warmth and empathy to patients and absorb the wisdom of human educators, should not be underestimated. The confusion about whether AI’s role in medical education is supplementary or replacement-based is another concern [47]. In summary, while AI promises great advances and changes in medicine, it also poses numerous challenges and problems. The medical community needs to proactively address these challenges, leverage AI technology benefits, and promote continuous innovation and improvement in medical services.

Figure 8 shows that big data has a significant intensity of 2.01, firmly at the top of the list, and has become the focus of medical education in the past 3 years. The emergence and proliferation of COVID-19 in 2019 ushered in the big data epoch in medicine, with telemedicine systems, clinical intelligent decision-making, and management systems taking on pivotal roles.

First, the advent of big data has catalyzed the innovation of medical teaching paradigms: what does the future hold for medical education in the digital age? A study conducted by Han et al zeroes in on a future medical education model that leans heavily on big data, cutting-edge technology, and AI, with the aim to cultivate a new breed of medical students who display enhanced humanistic attributes, co-operation capacity, patient-needs sensitivity, and societal and global orientation [46].

Second, big data has stimulated innovation in clinical medicine models: the integration of advanced technologies like machine learning, clinical intelligent decision and management systems, and electronic medical records has propelled shifts, innovation, and advancement within clinical medicine paradigms. The study by Kolachalama and Garg posits that AI, fueled by machine learning algorithms, is an emerging computer science branch that is swiftly gaining traction in health care. AI is anticipated to play an instrumental role in precision medicine and health [15]. In 2022, Watson and Wilkinson released a paper entitled “Digital Healthcare in COPD Management: A Narrative Review on the Advantages, Pitfalls, and Need for Further Research,” illustrating the vast potential of digital health care innovation [48]. During the COVID-19 pandemic, it was expected that big data would mitigate the workload for doctors interpreting digital data, enhance their diagnostic and prognostic abilities, equip clinicians with intelligent decision-making and management systems, and offer patients optimal clinical care and self-management strategies.

Undeniably, big data, akin to many emergent tools, is a double-edged sword. Ensuring its tailored use and dialectical treatment constitutes a crucial aspect of digital health, striving to exploit its merits while circumventing its demerits. The pursuit of enduring, comprehensive, and precise population health data management emerges as a long-term strategy.

The recent surge in terms indicates that “management” is intimately linked to “big data.” Confronting the colossal medical data of today, the incorporation of AI technology can enhance management efficiency in spheres, such as hospital medical management, disease surgery management, and chronic disease management, among others. AI algorithms are used to scrutinize data pertaining to patients’ hospitalization duration, hospitalization route, and climatic and temporal factors, which effectively curtail the hospitalization duration and significantly rectify issues, such as the misallocation of medical resources [49]. Leveraging a diabetic retinopathy automatic grading and training system furnished with an AI-driven diagnosis algorithm to groom budding doctors can augment diagnostic accuracy, thereby strengthening DR management [50]. Surgical video, a crucial data source for medical education, should be systematically stored and managed. A system intended to assist doctors in managing surgical videos can heighten the efficiency of continuing education by dissecting surgical videos and marking critical segments or frames to generate AI reports [51].