{"title":"Pedagogical Efficacy of LLM-Generated Synthetic Data Versus Real-World Clinical Records: A Randomized Controlled Non-Inferiority Trial.","authors":"Zhongkun Zuo, Jianyu Fang, Muyao Ye, Zedong Li","doi":"10.5334/pme.2535","DOIUrl":null,"url":null,"abstract":"<p><strong>Background: </strong>Expert-reviewed clinical cases generated by large language models (LLMs) may supplement case resources in medical education, but their short-term educational performance relative to real-case-derived teaching materials remains uncertain. We compared immediate post-training test performance after teaching with the two types of case materials and assessed non-inferiority against a prespecified margin.</p><p><strong>Methods: </strong>We conducted a prospective, parallel-group, randomized non-inferiority trial. Through the Wenjuanxing online platform, participants were randomized 1:1 to learn with either real-case-derived teaching cases compiled by clinicians and reviewed by experts or AI-generated clinical cases produced by Gemini 3.0 Pro from fully de-identified matched real cases and reviewed by three senior general surgery specialists with full-professor rank. The primary outcome was the total score on an independent 10-item immediate post-training test (0-10 points), with a prespecified non-inferiority margin of -0.5 points. Secondary outcomes included the training-phase performance score, learning efficiency index, single-item mental effort rating, case realism, and case-source judgment.</p><p><strong>Results: </strong>A total of 403 participants were randomized, of whom 386 were included in the modified intention-to-treat analysis: 192 in the real-case group and 194 in the AI-generated case group. The mean post-training test score was 4.95 (SD, 3.35) in the real-case group and 4.61 (SD, 3.35) in the AI-generated case group. The mean difference (AI-generated minus real-case group) was -0.335 points (95% CI, -1.006 to 0.337). Because the lower bound of the confidence interval was below the prespecified non-inferiority margin of -0.5 points, non-inferiority was not demonstrated (one-sided P = 0.314). No significant between-group differences were observed in the training-phase performance score, learning efficiency index, or single-item mental effort rating. AI-generated cases received lower realism ratings for Level 3 cases. The proportion of participants with at least one high-confidence completely incorrect response was 1.6% in the real-case group and 2.1% in the AI-generated case group.</p><p><strong>Conclusions: </strong>In this short-term, text-based online case-learning setting, no statistically significant between-group difference was observed in immediate post-training test performance; however, non-inferiority of AI-generated clinical cases relative to real-case-derived teaching materials was not demonstrated.</p>","PeriodicalId":48532,"journal":{"name":"Perspectives on Medical Education","volume":"15 1","pages":"759-770"},"PeriodicalIF":3.5000,"publicationDate":"2026-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13564930/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Perspectives on Medical Education","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.5334/pme.2535","RegionNum":2,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2026/1/1 0:00:00","PubModel":"eCollection","JCR":"Q1","JCRName":"EDUCATION, SCIENTIFIC DISCIPLINES","Score":null,"Total":0}
引用次数: 0
Abstract
Background: Expert-reviewed clinical cases generated by large language models (LLMs) may supplement case resources in medical education, but their short-term educational performance relative to real-case-derived teaching materials remains uncertain. We compared immediate post-training test performance after teaching with the two types of case materials and assessed non-inferiority against a prespecified margin.
Methods: We conducted a prospective, parallel-group, randomized non-inferiority trial. Through the Wenjuanxing online platform, participants were randomized 1:1 to learn with either real-case-derived teaching cases compiled by clinicians and reviewed by experts or AI-generated clinical cases produced by Gemini 3.0 Pro from fully de-identified matched real cases and reviewed by three senior general surgery specialists with full-professor rank. The primary outcome was the total score on an independent 10-item immediate post-training test (0-10 points), with a prespecified non-inferiority margin of -0.5 points. Secondary outcomes included the training-phase performance score, learning efficiency index, single-item mental effort rating, case realism, and case-source judgment.
Results: A total of 403 participants were randomized, of whom 386 were included in the modified intention-to-treat analysis: 192 in the real-case group and 194 in the AI-generated case group. The mean post-training test score was 4.95 (SD, 3.35) in the real-case group and 4.61 (SD, 3.35) in the AI-generated case group. The mean difference (AI-generated minus real-case group) was -0.335 points (95% CI, -1.006 to 0.337). Because the lower bound of the confidence interval was below the prespecified non-inferiority margin of -0.5 points, non-inferiority was not demonstrated (one-sided P = 0.314). No significant between-group differences were observed in the training-phase performance score, learning efficiency index, or single-item mental effort rating. AI-generated cases received lower realism ratings for Level 3 cases. The proportion of participants with at least one high-confidence completely incorrect response was 1.6% in the real-case group and 2.1% in the AI-generated case group.
Conclusions: In this short-term, text-based online case-learning setting, no statistically significant between-group difference was observed in immediate post-training test performance; however, non-inferiority of AI-generated clinical cases relative to real-case-derived teaching materials was not demonstrated.
期刊介绍：
Perspectives on Medical Education mission is support and enrich collaborative scholarship between education researchers and clinical educators, and to advance new knowledge regarding clinical education practices.
Official journal of the The Netherlands Association of Medical Education (NVMO).
Perspectives on Medical Education is a non-profit Open Access journal with no charges for authors to submit or publish an article, and the full text of all articles is freely available immediately upon publication, thanks to the sponsorship of The Netherlands Association for Medical Education.
Perspectives on Medical Education is highly visible thanks to its unrestricted online access policy.
Perspectives on Medical Education positions itself at the dynamic intersection of educational research and clinical education. While other journals in the health professional education domain orient predominantly to education researchers or to clinical educators, Perspectives positions itself at the collaborative interface between these perspectives. This unique positioning reflects the journal’s mission to support and enrich collaborative scholarship between education researchers and clinical educators, and to advance new knowledge regarding clinical education practices. Reflecting this mission, the journal both welcomes original research papers arising from scholarly collaborations among clinicians, teachers and researchers and papers providing resources to develop the community’s ability to conduct such collaborative research. The journal’s audience includes researchers and practitioners: researchers who wish to explore challenging questions of health professions education and clinical teachers who wish to both advance their practice and envision for themselves a collaborative role in scholarly educational innovation. This audience of researchers, clinicians and educators is both international and interdisciplinary.
The journal has a long history. In 1982, the journal was founded by the Dutch Association for Medical Education, as a Dutch language journal (Netherlands Journal of Medical Education). As a Dutch journal it fuelled educational research and innovation in the Netherlands. It is one of the factors for the Dutch success in medical education. In 2012, it widened its scope, transforming into an international English language journal. The journal swiftly became international in all aspects: the readers, authors, reviewers and editorial board members.
The editorial board members represent the different parental disciplines in the field of medical education, e.g. clinicians, social scientists, biomedical scientists, statisticians and linguists. Several of them are leading scholars. Three of the editors are in the top ten of most cited authors in the medical education field. Two editors were awarded the Karolinska Institute Prize for Research. Presently, Erik Driessen leads the journal as Editor in Chief.
Perspectives on Medical Education is highly visible thanks to its unrestricted online access policy. It is sponsored by theThe Netherlands Association of Medical Education and offers free manuscript submission.
Perspectives on Medical Education positions itself at the dynamic intersection of educational research and clinical education. While other journals in the health professional education domain orient predominantly to education researchers or to clinical educators, Perspectives positions itself at the collaborative interface between these perspectives. This unique positioning reflects the journal’s mission to support and enrich collaborative scholarship between education researchers and clinical educators, and to advance new knowledge regarding clinical education practices. Reflecting this mission, the journal both welcomes original research papers arising from scholarly collaborations among clinicians, teachers and researchers and papers providing resources to develop the community’s ability to conduct such collaborative research. The journal’s audience includes researchers and practitioners: researchers who wish to explore challenging questions of health professions education and clinical teachers who wish to both advance their practice and envision for themselves a collaborative role in scholarly educational innovation. This audience of researchers, clinicians and educators is both international and interdisciplinary.
The journal has a long history. In 1982, the journal was founded by the Dutch Association for Medical Education, as a Dutch language journal (Netherlands Journal of Medical Education). As a Dutch journal it fuelled educational research and innovation in the Netherlands. It is one of the factors for the Dutch success in medical education. In 2012, it widened its scope, transforming into an international English language journal. The journal swiftly became international in all aspects: the readers, authors, reviewers and editorial board members.
The editorial board members represent the different parental disciplines in the field of medical education, e.g. clinicians, social scientists, biomedical scientists, statisticians and linguists. Several of them are leading scholars. Three of the editors are in the top ten of most cited authors in the medical education field. Two editors were awarded the Karolinska Institute Prize for Research. Presently, Erik Driessen leads the journal as Editor in Chief.
Perspectives on Medical Education is highly visible thanks to its unrestricted online access policy. It is sponsored by theThe Netherlands Association of Medical Education and offers free manuscript submission.
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号
