How much MedEd AI research actually happening?

Four new bibliometric studies map the literature itself, and the picture is bigger, faster and shakier than any single paper can show.

This Week in MedEd

1 new event, 23 jobs (including 8 new and 2 closing this week), including:

• 💰 Advanced Digital Skills for AI Uptake in Health — up to €3.9M, EU consortia, deadline 1 Oct 2026

• 🎓 EBMA 2026: Annual European Conference on Assessment in Medical Education — Lancaster, UK, 1–3 Dec 2026

• 💼 Associate Dean of Digital Education — BIMM University, London (closes 18 Aug)

The full list, with deadlines, links and eligibility, is one tap away. Try everything free for 30 days.

Every week I sift through another dozen papers, each reporting on its own small corner of AI in medical education: one study, one cohort, one tool. It is easy to lose sight of how large this literature has become, or where it is actually heading. This week I am setting the individual studies aside and looking instead at four papers that undertake bibliometric analysis of the field itself.

I am covering four papers this week. One that analyses the whole of medical education research, then AI in medical education specifically, then large language models alone, and finally within a single specialty (anatomy). They show a field that is expanding fast and concentrated in a handful of countries and journals.

You can read about how I use AI to keep on top of the literature here:

Key points

• Across the whole of medical education research genAI is among only 10 of 38 identified topics whose share of publications is significantly rising.

• There were 852 papers specifically on AI in medical education in 2025 alone, more than were published across the entire decade to 2020.

• A subset of that literature focused on LLMs and is growing even faster (60% increase per year) , but it remains heavy on opinion and framework pieces relative to controlled trials.

• The United States and China dominate authorship, institutions, and citation impact.

Chen et al. applying BERTopic, a transformer-based topic model, to 276,253 English-language medical education publications from PubMed, Web of Science, Scopus, and OpenAlex between 2000 and 2024¹. They identified 38 substantive topics, 10 of which showed a significantly rising share of output, including AI and ChatGPT alongside virtual reality simulation, burnout and wellbeing, and climate and health education.

Twelve topics were cooling and sixteen showed no change.

The principle finding is that AI is not medical education's only hot topic, it is one of ten, rising alongside equity, wellbeing, and simulation rather than crowding them out. For anyone trying to guess where curriculum time and research funding will go next, this is a useful corrective to AI-only tunnel vision.

Ling and Wang focus on AI in medical education specifically, screening 3,474 Web of Science records down to 1,853 core publications spanning 2015-2025². Annual output has climbed every year; there were 852 papers published in 2025, which is more than the cumulative total for 2015 to 2020. The United States and China supply most of the leading authors and institutions, and the core publishing venues are BMC Medical Education, the Journal of Medical Internet Research (JMIR), and Medical Teacher.

The authors identified three research hotspots, educational scenarios and participants, core AI technologies, and educational programs, and argue the field has now moved beyond simple feasibility studies into what they call ‘deep integration’, where machine learning, deep learning, and large language models are combined rather than tested in isolation.

Huang and Liu zoom in further still by focussing only on the LLM literature³: 1,991 papers published between 2016 and March 2026, growing at an annual rate of 59%. JMIR Medical Education emerges as the field's core journal, and the authors identify five themes:

1. Education ethics

2. LLM performance

3. Patient education

4. Clinical reasoning

5. Intelligent assessment

Their conclusion is cautionary: research priorities have shifted from basic neural network exploration toward privacy and deployment questions, but empirical evidence has not kept pace with the volume of opinion and framework papers, and governance frameworks are lagging behind adoption.

They recommend more cohort trials, specialty-specific tools, and unified ethical standards. The take-home message here is as we have discussed in this newsletter before: the literature is still dominated by conceptual frameworks and opinion pieces.

Fan et al. have undertaken a bibliometric analysis of AI in anatomy education⁴. Their review covers 184 papers published between 2005 and 2024, again with the United States leading on output and citation impact. Their findings mirror the broader: early work concentrated on general educational applications, then shifted toward AI-driven image segmentation, deep learning, and surgical guidance, with large language models such as ChatGPT only recently starting to attract attention.

It is a useful reminder that the "field" these bibliometric tools describe is really dozens of smaller fields moving through the same cycle at different speeds.

None of these four papers ran a trial or surveyed a single student, and that is precisely what makes them useful. They confirm that the growth in AI Medical Education research is real and accelerating , not just a subjective impression from an overflowing inbox, and they locate where the genuine gaps are: empirical evidence lagging behind framework papers and a handful of countries doing most of the publishing. If you are deciding where to focus your own reading or research this year, these maps are a reasonable place to start, and if you struggling to keep on top of all the AI MedEd papers, you can see every single one that has been written about in this newsletter here.

1. Chen C, Li J, Zhang Z, Zhang T. Thematic Evolution in Medical Education Research (2000–2024): A Large-Scale BERTopic Analysis. Advances in Medical Education and Practice. 2026. https://doi.org/10.2147/AMEP.S618140

2. Ling X, Wang C. Research landscape thematic evolution and future trends of artificial intelligence in medical education from 2015 to 2025. Discover Computing. 2026;29:447. https://doi.org/10.1007/s10791-026-10349-w

3. Huang C, Liu W. Hotspot Evolution and Future Prospects of Large Language Models in Medical Education: A Bibliometric Analysis. Advances in Medical Education and Practice. 2026. https://doi.org/10.2147/AMEP.S625578

4. Fan Y, Zheng X, Yang G, Li P, Ran Y, Xu T, Wei T. A bibliometric analysis of artificial intelligence in anatomy education: Current situation, hot spots, and global trends. Medicine. 2026. https://journals.lww.com/md-journal/fulltext/2026/05290/a_bibliometric_analysis_of_artificial_intelligence.99.aspx