Purpose
This study explores the development and impact of DDx, an AI-driven medical case-based learning platform designed to simulate realistic patient interactions. Our goal was to address limitations of traditional standardized patient (SP) simulations, providing an accessible, scalable, and cost-effective alternative that enhances clinical training.

Methods
DDx leverages advanced large language models (LLMs) to create dynamic and responsive virtual patients, engaging learners in realistic diagnostic conversations. Piloted at Charles R. Drew University with 60 first-year medical students, three cases (STEMI, PE, AAA) were implemented in small group settings. Students completed cases in class and accessed them on demand for reinforcement. Survey feedback informed iterative improvements, including UX/UI enhancements, model adjustments, and scoring refinements.

Results
In a pilot study at Charles Drew University, 60 students engaged in 1,422 clinical interactions and reported high effectiveness and engagement. Survey results showed a 48% overall improvement in perceived clinical reasoning skills, with notable increases in comfort levels for history taking (+30%), interpreting physical exam findings (+53%), and developing differential diagnoses (+47%). These results highlight the potential of AIVPs to enhance medical training significantly. This demonstration will showcase DDx, highlighting our design principles, implementation challenges, and the impact on medical education.

Conclusion
AI-powered virtual patients, as exemplified by DDx, offer a transformative approach to medical education. By reducing costs, increasing accessibility, and enabling repeated practice, DDx addresses key limitations of traditional SP simulations. This innovative platform holds promise for advancing diagnostic reasoning and clinical preparedness in medical training.