John Taylor, Professor of Economics at Stanford University and developer of the "Taylor Rule" for setting interest rates | Stanford University
John Taylor, Professor of Economics at Stanford University and developer of the "Taylor Rule" for setting interest rates | Stanford University
Marcos Rojas, a doctoral student at Stanford Graduate School of Education (GSE), is piloting a web-based platform designed to enhance medical students' diagnostic skills. The tool, named Clinical Mind AI, uses generative artificial intelligence (AI) to simulate real-world interactions between physicians and patients, providing feedback on the exchange and the validity of diagnoses.
“The best way to learn about clinical reasoning, other than with actual patients, is through in-person simulations, with trained actors who are hired to act out scenarios,” said Rojas. He noted that such simulations are costly and not scalable. “With AI, I saw a way to overcome these challenges, to create a cost-effective, scalable tool.”
The platform was developed in the IDEAL Research Lab at Stanford, which focuses on addressing barriers to equity in learning environments. GSE Assistant Professor Shima Salehi directs the lab and serves as co-principal investigator on the project alongside Thomas Caruso from the School of Medicine.
Rojas shifted his research focus after the launch of ChatGPT in late 2022. He enrolled in a computer science course on AI applications in healthcare and created an initial prototype for Clinical Mind AI during this time.
Existing methods for teaching clinical reasoning often involve written summaries of clinical cases or preset case studies that may not be culturally relevant or flexible enough for different theoretical frameworks. “Different instructors understand clinical reasoning and learning outcomes in different ways,” Rojas explained. “These platforms are just too rigid in language, content, and understanding of the cognitive task that is being carried out.”
Clinical Mind AI allows educators to input their own content and demographic profiles while enabling interactions in various languages. The chatbot provides customizable feedback based on students' performance during diagnostic exercises.
Salehi highlighted the project's broader implications: “At the IDEAL Research Lab, we want everyone to have equal access to fields that can be engines for social mobility.” She praised Rojas' project for its applicability across cultures and languages without requiring significant resources from institutions.
Caruso emphasized the collaboration's success due to its mission-focused approach: “In interdisciplinary research, people sometimes lose sight of the mission... But this collaboration worked because we’re all mission-focused.”
Rojas initially tested Clinical Mind AI with a focus group comprising medical students, instructors, and attending physicians. He is currently testing it with 100 medical instructors across the United States and plans international trials in 2025.
Rojas stressed his commitment to assessing the tool's effectiveness through rigorous research: “We need evidence...to make sure it serves an educational purpose.” The project has received support from various Stanford entities including HAI and Impact Labs.
Future plans include expanding Clinical Mind AI’s capabilities to conduct physical exams and medical tests while incorporating more metrics for skill assessment. The team aims to offer the tool at no cost to resource-limited institutions worldwide.
###
Alerts Sign-up