Our AI system is built upon a secure, institutionally approved, infrastructure using OpenAI’s real-time voice model within an academic medical center’s Azure AI Foundry environment. Learners engage the AI through natural spoken dialogue, navigating realistic patient scenarios in diverse settings including outpatient clinics, emergency departments, operating rooms, and intensive care units. 

The AI prompts participants to reason through diagnostic and management decisions, mirroring the flow of an oral examination, and providing structured, domain-specific feedback on patient care, medical knowledge, and decision-making. No new AI model development is performed under this study; rather, the research evaluates the educational use case of an existing, approved platform. 

Pilot implementation has demonstrated adaptability across numerous topic domains including gastroenterology, critical care, internal medicine, and surgical decision-making. The system records de-identified interaction metadata—such as duration, message count, and learning objectives—to assess engagement and feasibility. 

Preliminary Results 

Initial user testing sessions (n=20) showed: 

All sessions were conducted in secure institutional environments, with no patient data used or retained. 

Conclusions and Future Directions 

This technology demonstrates a novel application of established AI infrastructure for immersive, oral board–style training across specialties. It represents a feasible and scalable educational intervention that aligns with accepted practices in simulation-based medical education.