BERTopic analysis of 276,000 papers maps hot and cold trends in medical education research (2000–2024)

Among 38 substantive topics, 10 showed increasing relative shares ('hot'), including AI/ChatGPT, VR simulation, burnout/wellbeing, diversity/equity, empathy, and climate/health education. Twelve topics declined ('cold'), while 16 remained stable. Findings highlight a gradual shi…

• Transformer-based BERTopic model analyzed 276,253 English-language publications (2000–2024) from PubMed, Web of Science, Scopus, and OpenAlex.
• Topic trends classified via Poisson regression with false discovery rate correction; 35 of 38 trend classifications agreed across alternative standard-error specifications.
• Retrospective bibliometric design; temporal associations with policy/technology events are exploratory and not causal (exact figure not reported in source).

RadPigeon summaries are original and for information only. They are not clinical advice.