A French OSCE Dialogue Dataset and Controllable Virtual Patient System for Clinical Training

2026-06-26

AI Summary

This study addresses the scarcity of standardized patients in medical education, which limits effective training in clinical communication skills. To overcome this challenge, the authors construct a French-language OSCE dialogue dataset comprising 240 student–patient interactions and propose a modular, controllable generation framework that integrates retrieval-augmented generation with a reflection loop mechanism to develop a high-fidelity virtual patient system. The system enables realistic and coherent clinical scenario simulations with automated feedback and incorporates a multi-level evaluation framework alongside an LLM-as-a-Judge approach to significantly enhance both the authenticity of virtual patients and the consistency of student assessments. Experimental results validate the effectiveness of the proposed methodology, and a functional prototype has been implemented to support interactive practice with real-time feedback.

Abstract

The clinical and communication skills of medical students are commonly assessed through Objective Structured Clinical Examinations (OSCEs), which consist of brief scenario-driven simulations of doctor-patient interactions. However, training is often limited by the low availability of human standardized patients, motivating the development of realistic virtual patients (VPs). To address this gap, we introduce a French OSCE dialogue dataset comprising 240 student-patient training interactions. We build upon it a controllable LLM-based pipeline to generate synthetic OSCE dialogues. The pipeline integrates modular components, such as retrieval-based grounding and a reflection loop, to ensure patient fidelity, coherence, and realism. Additionally, we propose a multi-level evaluation framework assessing patient simulation quality, student performance, and linguistic quality, using an LLM-as-a-Judge approach. Experiments suggest that controllability modules generally improve patient fidelity and student evaluation consistency. Finally, we implement an interactive prototype in which students can practice with a VP and receive automatic feedback.

Problem

• OSCE
• virtual patient
• clinical training
• standardized patient
• medical education

Innovation

• controllable LLM
• virtual patient
• OSCE dialogue dataset
• retrieval-based grounding
• LLM-as-a-Judge