Projects

We explore how AI enhances clinical reasoning, focusing on AI-simulated patient interactions, usability, and global impact. Through innovative research and global collaboration, we aim to transform medical education and improve healthcare outcomes.

We are currently focused on the following research projects:

• Feasibility and Acceptability of the Clinical Mind AI Platform for Medical Instructors
  This research explores how medical instructors can use an AI-powered platform to simulate realistic patient interactions for teaching clinical reasoning. The study focuses on whether the AI can accurately simulate patients based on instructor-provided clinical cases, what improvements are needed for the platform’s interface, potential barriers to adoption in medical education, and what insights instructors need to assess student performance. Through a mix of usability questionnaires, interviews, and analysis of AI-generated simulations, the research aims to ensure the tool’s feasibility, acceptability, and effectiveness in enhancing clinical reasoning education.
• Assessment of Clinical Reasoning Skills using AI-Simulated Patients: Initial Validity Evidence of the Platform Clinical Mind AI
  This research aims to explore the use of AI-simulated patients to assess clinical reasoning skills in medical students, focusing on the critical phase of gathering patient history. It proposes evaluating how students identify key case features, apply predictive frameworks, and consider potential diagnoses after interacting with an AI-simulated patient. Using a targeted scoring rubric and large language models for automated assessment, the study seeks to establish the validity of this approach, comparing AI and human scoring to ensure reliability and accuracy in evaluating clinical reasoning. The research serves as a proof of concept for scalable, AI-driven assessment tools in medical education.
• International Pilot of AI-Simulated Patients for Clinical Reasoning Education
  This international pilot study evaluates the feasibility and adaptability of AI-simulated patients for teaching clinical reasoning in several countries. Medical instructors will create clinical case scripts to generate AI-simulated patients tailored to their curricula and healthcare contexts. Medical students will engage with these simulations to gather patient histories and develop diagnoses. The study assesses usability, simulation accuracy, and alignment with local medical practices, providing critical insights to refine and scale this AI-driven educational tool globally.
• Minimum Features for Generating AI-Simulated Patient Headshots: A Cross-Regional Analysis of Medical Instructor Satisfaction
  This pilot study investigates the essential patient characteristics required to generate realistic AI-generated headshots that align with medical instructors' intended representations of simulated patients. Instructors from diverse global regions will create clinical case scripts on the Clinical Mind AI platform, with headshots generated using three AI tools: Midjourney, Gemini, and ChatGPT DALL-E 3. Satisfaction levels will be assessed through Likert scales and qualitative feedback, alongside analysis of regional preferences and potential biases. The study also evaluates cost-effectiveness across AI tools to optimize this feature for enhancing medical simulations worldwide.
• Conversational Artificial Intelligence Simulations to Improve Non-Technical Skills Among Anesthesiology Trainees: A Prospective, Mixed-Methods Study
  This study evaluates the use of conversational AI simulations as a tool for training anesthesiology trainees in managing difficult conversations, such as disclosing medical errors to patients' families. Trainees interact with an AI-simulated scenario, practicing communication skills based on established frameworks like SPIKES and the Anesthetists’ Non-Technical Skills (ANTS) framework. The AI also provides automated feedback on performance, which is compared to expert human assessments. The study explores the acceptability, usability, and effectiveness of this innovative approach in enhancing non-technical skills, with the goal of offering a scalable, cost-effective alternative to traditional simulation methods.