Purpose
Effective communication during challenging patient interactions is essential for medical professionals. Simulation-based learning is effective but resource-intensive, so medical students often lack scalable, accessible training opportunities. To address this gap, we developed a novel educational module combining instructional videos and an AI-powered patient simulation app. This tool provides preclinical students with an immersive, scalable platform to practice delivering difficult news.

Methods
The educational module was integrated into the curriculum and tested with 96 second-year medical students in the "On Doctoring" course at the Geisel School of Medicine at Dartmouth. Video tutorials on the SPIKES protocol for delivering bad news were combined with interactive communication training via an AI-based patient simulation app. The app, openly accessible at https://patient-actor-dc.streamlit.app/, employs a Large Language Model (LLM) powered by GPT-4 to simulate realistic patient interactions across diverse clinical scenarios, including emotional and contextual cues and patient personality profiles. Students received immediate feedback based on SPIKES criteria from an AI-driven evaluator embedded in the app. Student attitudes toward the AI Patient Actor were assessed, including perceptions of the chatbot's realistic conversational abilities and the usefulness of feedback.

Results
Findings show that the AI Patient Actor provides realistic patient responses in a difficult conversation context, as well as meaningful feedback for improving communication skills and adherence to the SPIKES model. We present a comprehensive summary of results from the mixed-methods analysis, including qualitative survey responses and sentiment analysis of open-text feedback about student attitudes, beliefs, and preferences towards conversational AI in clinical skills education.

Conclusion
This innovative educational module addresses a critical gap in healthcare education by providing an accessible, scalable tool for communication skills training that can enhance student confidence and competency in navigating difficult conversations. Future validation studies should evaluate its impact on communication skills and clinical reasoning in healthcare trainees.