AI-Enabled Virtual Patients as Part of Clinical Skills Training: A Cross-Sectional Program Evaluation of Student Perspectives on the McMaster Virtual SP Tool
Cross-sectional evaluation reveals high acceptability of generative AI virtual patients among medical students, suggesting value as adjuncts for cognitive clinical rehearsal.
Key Points
To evaluate medical students' awareness, utilization, and perceptions of a generative artificial intelligence-enabled virtual standardized patient tool designed to supplement clinical skills practice.
Conducted an anonymous, single-institution cross-sectional survey across three medical student cohorts (N=35).
Summarized quantitative metrics descriptively and analyzed free-text responses using qualitative content analysis framed by deliberate practice and simulation instructional design principles.
Overall, 23 of 35 respondents (65.7%) were aware of the tool, 16 (45.7%) had used it, and 14 (40.0%) utilized AI-enabled virtual patients for OSCE preparation.
Among users (n=16), 15 (93.8%) found navigation easy, 13 (81.3%) intended to reuse the tool, and 13 (81.3%) would recommend it.
Qualitative analysis identified strengths in accessibility and low-stakes history-taking rehearsal, balanced by limitations in emotional realism, nonverbal cues, physical examination practice, and feedback specificity.