DIALOGUE: A Generative AI–Based Pre–Post Simulation Study to Improve Diagnostic Communication in Medical Students Using Type 2 Diabetes Scenarios

Effective diagnostic communication—delivering a diagnosis with clarity, structure, and empathy—remains a challenging competency for many undergraduate medical students. This single-arm pre–post study evaluated a generative artificial-intelligence (GenAI) training module designed to improve diagnostic-communication performance. Thirty clinical-phase students completed two pre-test encounters in which they disclosed a type 2 diabetes mellitus (T2DM) diagnosis to a virtual patient powered by ChatGPT (GPT-4o) and were scored with an eight-domain rubric by blinded raters. They then undertook ten asynchronous GenAI scenarios with automated natural-language feedback, followed seven days later by two post-test consultations with human standardized patients assessed in real time with the same rubric. Mean total performance increased by 36.7 points (95 % CI: 31.4–42.1; p < 0.001), and the proportion of high-performing students rose from 0 % to 70 %. Gains were significant across all domains, most notably in opening the encounter, closure, and diabetes-specific explanation. Multiple regression showed that lower baseline empathy (β = –0.41, p= 0.005) and higher digital self-efficacy (β = 0.35, p= 0.016) inde-pendently predicted greater improvement; gender displayed only a marginal effect. Cluster analysis revealed three learner profiles, with the highest-gain cluster character-ised by low empathy and high digital self-efficacy. Inter-rater reliability was excellent (ICC ≈ 0.90). These findings provide empirical evidence that GenAI-mediated practice can produce meaningful, measurable enhancements in diagnostic-communication skills and may serve as a scalable, individualised adjunct to conventional clinical education.