Combining generative artificial intelligence practice with standardized patient assessment to develop medical students' history taking and physical examination skills: an educational case report

This dataset contains the anonymized, participant-level data underlying an educational case report on a hybrid clinical-skills training intervention. Twenty-two third-year medical students at a single European medical school completed a four-week intervention combining two generative-AI tools (a virtual patient for history taking and a Socratic tutor for physical examination) with three standardized clinical assessments conducted by standardized patients. The file includes, for each participant, initial-tool group allocation, sex, analytic rubric scores for history taking (0–9, with communication, structure, and clinical-reasoning subdomains) and physical examination (0–7, with communication/hygiene and technique subdomains) at weeks 0, 2, and 4, cumulative usage hours for each tool, and post-study satisfaction survey responses (5-point Likert). Direct identifiers have been removed and quasi-identifiers coarsened to prevent re-identification. Data are provided in SPSS (.sav) and CSV (.csv) formats, accompanied by a variable codebook. These data support the analyses reported in the associated manuscript and are shared to enable verification and reuse. Any secondary analysis should respect the original consent scope; the external comparison-group grades reported in the manuscript are not individually redistributable and are therefore not included. Keywords: medical education; generative artificial intelligence; clinical skills; history taking; physical examination; simulated patients License: Creative Commons Attribution 4.0 (CC BY 4.0) — or, if using controlled access, Zenodo "Restricted Access." Related identifier: "is supplement to" → [manuscript DOI, once assigned]