Introduction

Background

Artificial intelligence (AI) is increasingly acknowledged as a transformative force in healthcare, with applications ranging from diagnostic support and clinical decision-making to personalized treatment [1]. Advances in machine learning, natural language processing, and computer vision enable AI systems to process vast datasets, discern complex patterns, and improve both the accuracy and efficiency of medical tasks. These innovations are reshaping not only healthcare delivery but also the training requirements for future professionals who must work effectively alongside AI technologies [2]. Traditional medical education, founded on evidence-based knowledge and humanistic principles, has emphasized clinical reasoning, anatomical understanding, and patient-centered care. Its core approaches have historically relied on mentorship, memorization, and clinical experience. However, these traditional methods face growing challenges in the context of rapid medical innovation and the ongoing digital transformation of healthcare. A significant gap has emerged between conventional training models and the digital competencies essential for modern clinical practice [3].

In response, AI is being increasingly incorporated into medical education. Technologies such as intelligent tutoring systems, virtual patients, and AI-assisted assessment tools facilitate more personalized, adaptive, and data-driven learning experiences. They also promote the development of skills needed for practice in AI-augmented clinical environments [4]. Despite the widespread adoption of AI in educational contexts, research on its impact remains fragmented, and a comprehensive overview of the field is lacking.

Objectives

To address this knowledge gap, bibliometric analysis provides a systematic approach for mapping the structure and evolution of research at the intersection of AI and medical education. By examining publication trends, citation patterns, and thematic developments, this study elucidates how AI is shaping medical education and identifies priority areas for future inquiry.

Methods

Ethics statement

This study did not involve human subjects; therefore, institutional review board approval and informed consent were not required.

Study design

A bibliometric and content analysis was conducted to systematically examine the landscape of AI applications in medical education. The methodological framework was based on established principles of scientometric research.

Eligibility criteria

Studies were eligible for inclusion if they were original research articles or reviews published in English. Exclusion criteria were as follows: duplicate records, non-English publications, conference proceedings, and articles unrelated to medical education or AI, as determined by title and abstract screening.

Information sources

Literature was sourced from the Scopus database, which is recognized for its comprehensive indexing of peer-reviewed publications across multiple disciplines.

Search strategy

The search encompassed publications available up to April 10, 2025. Search terms included “medical education” and AI-related phrases such as “artificial intelligence” and “AI.” Searches were conducted within the “Title,” “Abstract,” and “Keywords” fields. Filters were applied to include only “Article” and “Review” types, and to restrict the language to English.

Selection and data collection process

The initial search identified 2,112 records. To ensure rigor and relevance, two reviewers independently screened titles and abstracts, followed by full-text review when necessary. Following this screening and eligibility assessment, 732 articles were retained for final bibliometric analysis.

Data items

Extracted data included core bibliographic information such as title, authors, year of publication, affiliations, source title, citation counts, keywords, abstract, and the digital object identifier (DOI). Additional identifiers and access details were also collected for reference and verification purposes.

Bibliometric and content analysis

Bibliometric analysis was performed using VOSviewer ver. 1.6.20 (Centre for Science and Technology Studies, Leiden University), focusing on keyword co-occurrence, citation coupling, and co-authorship networks. For qualitative synthesis, thematic clusters generated from co-occurrence mapping were further examined to interpret research foci, patterns of evolution, and underlying conceptual frameworks.

Results

Distribution of main characteristics

Research period

From 1986 to 2025, publications addressing the application of AI in medical education have exhibited distinct phases of growth (Fig. 1). During the initial period (1986–2017), research activity remained limited, reflecting the exploratory stage of AI’s integration into medical education. A gradual increase in publication numbers began in 2018, followed by a marked acceleration in 2023 and 2024, with 134 and 328 publications, respectively, signifying a period of explosive growth. The data for 2025 indicate 149 publications, underscoring ongoing scholarly interest.

Distribution of authors

The author landscape in AI and medical education has become increasingly active in recent years, with most leading contributors publishing primarily around 2023 (Fig. 2). Authors such as Winkler-Schwartz, Ledwos, and Mirchi are notable for their high citation counts and total link strength, despite entering the field relatively recently. Co-authorship analyses reveal tightly interconnected research clusters, including scholars like Yilmaz and Del Maestro. Additionally, emerging researchers such as Gichoya, Trivedi, and Hashimoto demonstrate high normalized citation scores. Influential early contributors such as Wartman continue to shape the field. Thematic clusters among authors vary, with some focusing on simulation and assessment, while others emphasize AI algorithm development and clinical application.

Distribution of countries

The global research landscape for AI in medical education displays pronounced regional patterns (Fig. 3). The United States leads in both the volume of publications (255 papers) and citation count (n=7,397), followed by the United Kingdom, Canada, the Netherlands, and Germany. A notable surge in 2023 reflects heightened interest in generative AI. In 2024, countries such as Saudi Arabia and Jordan significantly increased their research output. Nations including Ireland, Singapore, and Israel, though contributing fewer publications, demonstrate high impact through citation metrics. The United States, United Kingdom, and Canada persist as central hubs for international research collaboration.

Thematic analysis

Keyword co-occurrence network

A total of 1,474 author keywords were extracted from the collected articles, focusing on those that appeared more than five times (Fig. 4). To improve cluster clarity and highlight key research themes, a minimum cluster size of 5 was adopted. This process identified 71 keywords with more than five occurrences, forming four clusters with a combined total link strength of 3,173.

Keyword frequency analysis

Analysis of keyword frequency revealed 25 terms occurring more than 10 times, with four—“artificial intelligence,” “medical education,” “ChatGPT,” and “large language model”—exceeding 100 mentions. These central keywords also anchor the clustering structure and define distinct thematic groupings. The total link strength of 3,173 across all keywords indicates a high level of interconnection. These results suggest that research in AI and medical education is concentrated around a core set of highly interrelated terms.

Average year of keywords co-occurrence

Examining keyword co-occurrence and average year of publication revealed that high-frequency keywords such as “artificial intelligence” (2023.54), “medical education” (2023.42), “ChatGPT” (2023.96), and “large language model” (2024.13) were primarily clustered in the period from 2023 to 2024. Recently emerging terms such as “generative AI” (2024.17), “chatbot” (2023.88), and “NLP” (2023.89) were also frequent. In contrast, earlier terms like “deep learning” (2022.76), “simulation” (2021.92), and “radiology” (2022.73) show previous waves of adoption. Terms including “surgical education,” “assessment” (2023.58), and “examination” (2023.81) reflect a recent shift toward AI-assisted evaluation and testing.

Content analysis based on keyword clustering

Content analysis of the keyword clusters reveals four major thematic clusters, each representing a unique but interconnected research direction (Table 1).

Cluster 1

Generative AI and the transformation of medical assessment. This cluster is centered on the transformative impact of generative AI, especially tools such as ChatGPT (OpenAI) and other large language models (LLMs), in reshaping assessment and evaluation in medical education.

Cluster 2

Sociocultural and ethical dimensions of AI adoption. The second cluster highlights the sociocultural, ethical, and systemic considerations associated with integrating AI into medical education.

Cluster 3

Immersive technologies and personalized learning environments. This theme reflects the integration of immersive, adaptive, and AI-driven learning tools into the modern design of medical education.

Cluster 4

Learner perceptions and adoption of AI in clinical education. The fourth cluster addresses the experiences and perspectives of learners as they engage with AI technologies in their training.

Central node in citation networks

Citation analysis was performed on 732 articles, of which 218 had been cited more than 10 times. Among these, 27 articles received over 100 citations (Fig. 5). The most cited and most strongly co-linked publication in the dataset is “Performance of ChatGPT on USMLE: potential for AI-assisted medical education using large language models” by Kung et al. [5], published in 2023 in PLOS Digital Health. This study demonstrated that ChatGPT was able to pass the United States Medical Licensing Exam (USMLE) with over 60% accuracy and provide coherent clinical explanations. The article highlights the potential of AI for test preparation and personalized learning, but also cautions about possible inaccuracies and issues of trust. The authors call for ethical oversight and validation before widespread adoption, establishing this paper as a foundational reference in the field of AI in medical education.

Discussion

Key results

From 1986 to 2025, research on AI in medical education progressed through distinct phases. Early research (1986–2017) was slow, with single-digit annual publications, primarily using expert systems. Starting in 2018, publications rose, peaking at 328 in 2024. This change was driven by advances in deep learning and the emergence of LLMs such as ChatGPT. The United States leads with 255 publications, followed by the United Kingdom and Canada. Emerging nations such as Saudi Arabia show increased activity. Authors including Winkler-Schwartz and Del Maestro dominate citations, reflecting collaborative networks. Keywords such as “ChatGPT” and “medical education” reflect the central role of AI in assessment and personalized learning. Ethical considerations and the adoption of immersive technologies further define this rapidly expanding and globally collaborative field.

Interpretation/comparison with previous studies

The rapid rise of AI in medical education, especially since 2023, mirrors both rapid technological progress and growing enthusiasm for transforming healthcare training. A significant transition has occurred—from exploratory investigations to impactful, real-world applications—fueled in large part by LLMs such as ChatGPT [6]. The surge in research output from 2023–2024 highlights AI’s increasing influence on assessment, personalized learning, and clinical decision-making simulations [7].

A key finding is the expanding role of AI in enhancing assessment. Research demonstrates that AI tools like ChatGPT can meet or surpass passing thresholds in high-stakes examinations such as the USMLE, supporting their integration into formative assessment and self-directed learning [8]. These models offer dynamic feedback and can simulate clinical reasoning, addressing some limitations of traditional, static assessments. Nonetheless, concerns persist regarding their ability to interpret complex, context-rich clinical scenarios [9]. Effective clinical decision-making frequently relies on understanding cultural, historical, and personal patient contexts—areas where AI performance remains limited. This highlights the critical need to validate AI outputs for clinical appropriateness and avoid excessive reliance on automated systems. Additionally, growing awareness of the potential for AI to perpetuate or amplify biases—especially in clinical training—emphasizes the necessity for ongoing critical evaluation [10].

Ethical and sociocultural considerations are receiving increasing attention as AI adoption widens. Major concerns include algorithmic bias, transparency in decision-making, and accountability when using AI tools for diagnosis and care planning [11]. In clinical settings, opaque AI-driven decisions can erode patient trust and jeopardize safety. Therefore, safeguards must be built into AI systems, with human oversight remaining fundamental to clinical judgment [12]. Scholars emphasize that medical education should not only teach technical AI skills but also instill a strong sense of ethical responsibility [13]. This involves cultivating critical AI literacy—enabling students to engage not only with how AI works, but also with the broader societal implications of its use [14]. Ethics in AI is not simply about avoiding harm, but about reinforcing human-centered care and ensuring that AI augments, rather than replaces, human clinicians [15]. Thus, ethics education should prioritize transparency, fairness, and professional accountability to foster responsible application of AI in medicine.

Simultaneously, immersive technologies such as virtual reality and augmented reality are increasingly being integrated with AI to create adaptive, high-fidelity learning environments. These simulations provide risk-free opportunities for hands-on practice—especially important in fields such as surgery, where repeated skill acquisition is essential [16]. AI further enhances these environments by personalizing learning, adjusting in real time to each learner’s pace and performance. For example, platforms can identify knowledge gaps and deliver targeted content and feedback, strengthening learning outcomes [17]. Nevertheless, challenges remain in scaling these systems, particularly in low-resource settings, and ensuring the contextual relevance of AI-driven simulations across diverse populations.

Another trend is the rise of interdisciplinary collaboration. The increasing number of co-authored publications indicates that integrating AI into medical education requires contributions from medicine, computer science, ethics, and education [18]. Researchers from these diverse fields are joining forces to address the multifaceted challenges AI presents to educational and healthcare systems [19]. Collaboration between developers and medical educators ensures that AI tools are both technically robust and pedagogically appropriate [20]. Engaging ethicists and social scientists also helps to tackle concerns related to fairness, bias, and accountability. Such interdisciplinary partnerships are essential for aligning AI implementation with clinical realities and educational objectives [21].

Overall, AI’s influence in medical education is growing rapidly, moving from early proof-of-concept research to practical, real-world applications. While AI holds significant promise for improving assessment, simulation, and individualized learning, it also introduces complex ethical and operational challenges. Addressing these issues will require ongoing interdisciplinary collaboration, rigorous validation, and a strong commitment to responsible innovation. Future research should not only refine AI’s technical capabilities but also prioritize integration strategies that ensure safety, fairness, and optimal learner development. With appropriate oversight and sustained collaboration, AI has the potential to reshape medical education into a more dynamic, personalized, and equitable system.

Limitations

This study has several limitations. It focuses exclusively on English-language literature, which may have excluded relevant research published in other languages. While the discussion addresses AI’s applications, it does not examine in detail the real-world challenges faced by educators and students. The integration of AI across various institutions and countries presents unique challenges that warrant further investigation. Additionally, the long-term effects of AI on academic performance, clinical skills, and educational outcomes remain largely unexplored, with most current findings still preliminary. Ethical concerns, such as algorithmic bias, data privacy, and the evolving role of human judgment, are not central to this study, but are crucial topics for future research to ensure the responsible implementation of AI in medical education.

Conclusions

Through bibliometric and content analysis methods, this study systematically identified and evaluated current trends and challenges in the application of AI in medical education. The findings underscore the growing significance of AI, particularly LLMs like ChatGPT, in personalizing learning, advancing clinical training, and delivering real-time feedback. The study also addresses critical challenges, including the need for accurate AI-based assessments and the integration of AI tools within established educational frameworks. Overall, the adoption of AI in medical education signals a transformative shift toward more adaptive, efficient, and data-driven learning processes.

Notes

Conflict of interest

No potential conflict of interest relevant to this article was reported.

Funding

The authors received no financial support for this article.

Data availability

Dataset file is available from the Harvard Dataverse at https://doi.org/10.7910/DVN/4CNTII.

Dataset 1. Research data used for the bibliometric analysis using VOSviewer.

kcse-373-dataset-1.csv

Supplementary materials

The authors did not provide any supplementary materials for this article.

Fig. 1.

Number of publications over the years.

Fig. 2.

The density visualization of co-authorship authors.

Fig. 3.

The network visualization of co-authorship countries.

Fig. 4.

The network visualization of co-occurrence-author keywords.

Fig. 5.

The network visualization of citation cited references.

References

• 1. Alum EU, Ugwu OP. Artificial intelligence in personalized medicine: transforming diagnosis and treatment. Discov Appl Sci 2025;7:193. https://doi.org/10.1007/s42452-025-06625-x. Article
• 2. Al-Antari MA. Advancements in artificial intelligence for medical computer-aided diagnosis. Diagnostics (Basel) 2024;14:1265. https://doi.org/10.3390/diagnostics14121265. ArticlePubMedPMC
• 3. Car J, Ong QC, Erlikh Fox T, et al. The digital health competencies in medical education framework: an international consensus statement based on a Delphi study. JAMA Netw Open 2025;8:e2453131. https://doi.org/10.1001/jamanetworkopen.2024.53131. ArticlePubMed
• 4. Duan S, Liu C, Rong T, Zhao Y, Liu B. Integrating AI in medical education: a comprehensive study of medical students’ attitudes, concerns, and behavioral intentions. BMC Med Educ 2025;25:599. https://doi.org/10.1186/s12909-025-07177-9. ArticlePubMedPMC
• 5. Kung TH, Cheatham M, Medenilla A, et al. Performance of ChatGPT on USMLE: potential for AI-assisted medical education using large language models. PLOS Digit Health 2023;2:e0000198. https://doi.org/10.1371/journal.pdig.0000198. ArticlePubMedPMC
• 6. Adarkwah MA, Badu SA, Osei EA, Adu-Gyamfi E, Odame J, Schneider K. ChatGPT in healthcare education: a double-edged sword of trends, challenges, and opportunities. Discov Educ 2025;4:14. https://doi.org/10.1007/s44217-024-00393-3. Article
• 7. Fatima SS, Sheikh NA, Osama A. Authentic assessment in medical education: exploring AI integration and student-as-partners collaboration. Postgrad Med J 2024;100:959-67.https://doi.org/10.1093/postmj/qgae088. ArticlePubMed
• 8. Mackey BP, Garabet R, Maule L, Tadesse A, Cross J, Weingarten M. Evaluating ChatGPT-4 in medical education: an assessment of subject exam performance reveals limitations in clinical curriculum support for students. Discov Artif Intell 2024;4:38. https://doi.org/10.1007/s44163-024-00135-2. Article
• 9. Chan KY, Yuen TH, Co M. Using ChatGPT for medical education: the technical perspective. BMC Med Educ 2025;25:201. https://doi.org/10.1186/s12909-025-06785-9. ArticlePubMedPMC
• 10. Currie G, Currie J, Anderson S, Hewis J. Gender bias in generative artificial intelligence text-to-image depiction of medical students. Health Educ J 2024;83:732-46.https://doi.org/10.1177/00178969241274621. Article
• 11. Alrashed FA, Ahmad T, Almurdi MM, et al. Incorporating technology adoption in medical education: a qualitative study of medical students’ perspectives. Adv Med Educ Pract 2024;15:615-25.https://doi.org/10.2147/AMEP.S464555. ArticlePubMedPMC
• 12. Turner L, Hashimoto DA, Vasisht S, Schaye V. Demystifying AI: current state and future role in medical education assessment. Acad Med 2024;99:S42-7.https://doi.org/10.1097/ACM.0000000000005598. Article
• 13. Boscardin CK, Gin B, Golde PB, Hauer KE. ChatGPT and generative artificial intelligence for medical education: potential impact and opportunity. Acad Med 2024;99:22-7.https://doi.org/10.1097/ACM.0000000000005439. ArticlePubMed
• 14. Mondal H. Ethical engagement with artificial intelligence in medical education. Adv Physiol Educ 2025;49:163-5.https://doi.org/10.1152/advan.00188.2024. ArticlePubMed
• 15. Okamoto S, Kataoka M, Itano M, Sawai T. AI-based medical ethics education: examining the potential of large language models as a tool for virtue cultivation. BMC Med Educ 2025;25:185. https://doi.org/10.1186/s12909-025-06801-y. ArticlePubMedPMC
• 16. Lang M, Ghandour S, Rikard B, et al. Medical extended reality for radiology education and training. J Am Coll Radiol 2024;21:1583-94.https://doi.org/10.1016/j.jacr.2024.05.006. ArticlePubMed
• 17. Pu J, Hong J, Yu Q, et al. Accuracy, satisfaction, and impact of custom GPT in acquiring clinical knowledge: potential for AI-assisted medical education. Med Teach; 2025 Feb. 2. [Epub]. https://doi.org/10.1080/0142159X.2025.2458808. Article
• 18. Guraya SY. Transforming simulation in healthcare to enhance interprofessional collaboration leveraging big data analytics and artificial intelligence. BMC Med Educ 2024;24:941. https://doi.org/10.1186/s12909-024-05916-y. ArticlePubMedPMC
• 19. Greengrass C. Navigating the AI revolution in medicine: adopting strategies for medical education. Med Sci Educ 2024;35:1055-61.https://doi.org/10.1007/s40670-024-02257-9. ArticlePubMedPMC
• 20. Li R, Liu G, Zhang M, et al. Integration of artificial intelligence in radiology education: a requirements survey and recommendations from faculty radiologists, residents, and medical students. BMC Med Educ 2025;25:380. https://doi.org/10.1186/s12909-025-06859-8. ArticlePubMedPMC
• 21. Franco D’Souza R, Mathew M, Mishra V, Surapaneni KM. Twelve tips for addressing ethical concerns in the implementation of artificial intelligence in medical education. Med Educ Online 2024;29:2330250. https://doi.org/10.1080/10872981.2024.2330250. ArticlePubMedPMC

Figure & Data

References

Citations

Citations to this article as recorded by