Related Experiment Video
Updated: Aug 6, 2026
05:04
Using Simulation Models to Train Clinicians in the Use of Point-of-Care Ultrasound
Published on: August 9, 2024
AI-Simulated Patients for Training Shared Decision-Making: Feasibility Study in Medical Education
Lea Herschbach1, Teresa Festl-Wietek1, Alessandra Sonanini1
1TIME - Tuebingen Institute for Medical Education, University of Tübingen, Tuebingen, Baden-Wurttemberg, Germany.
JMIR Medical Education
|July 17, 2026
Summary
AI chatbots show promise for training shared decision-making (SDM) skills, with high agreement between AI and human ratings of performance. Medical students outperformed physicians in SDM skills during AI simulations.
Area of Science:
- Medical Education Technology
- Artificial Intelligence in Healthcare
- Patient-Centered Care
Background:
- Shared decision-making (SDM) is crucial for patient-centered care but lacks scalable training methods.
- AI-powered chatbots offer potential for standardized SDM practice and automated feedback.
Purpose of the Study:
- To assess agreement between AI and human ratings of SDM performance using the OPTION-12 scale.
- To compare SDM performance and attitudes between medical students and physicians.
- To explore changes in communicative self-efficacy (CSE) after chatbot-based SDM practice.
Main Methods:
- Feasibility study with a pre-post mixed methods design involving medical students and physicians.
- Participants engaged in a 20-minute AI chatbot interaction simulating a patient encounter for SDM practice.
- SDM performance (OPTION-12) and CSE were assessed pre- and post-intervention; AI ratings were compared to human ratings.
Main Results:
- High agreement (ICC=0.86) was found between AI-generated and human OPTION-12 scores, though AI ratings were higher.
- Medical students demonstrated higher SDM performance scores than physicians; both groups had positive SDM attitudes.
- Communicative self-efficacy (CSE) slightly decreased post-intervention; participants found the AI interaction authentic and useful.
Conclusions:
- AI-simulated patients are a feasible tool for SDM training in medical education, validating the OPTION-12 scale in this context.
- The study suggests potential for AI chatbots to support communication skills development and provide feedback.
- Larger, controlled studies are needed to confirm effectiveness and generalizability due to study limitations.
