Integrating State-of-the-Art Personalized Knowledge and Technologies into Medical Training
Our Mission
Why personalization in interactions matters
Effective care is built in conversation. Patients differ in how they describe symptoms, what they fear or value, how they speak with clinicians, and how directly they request changes in treatment. Recognizing these personality-driven differences—and adapting communication accordingly—is essential for accurate understanding, adherence, and a strong professional identity in medicine. This focus grows out of our group’s long-standing research on personality in social interactions and its impact on real-world behavior.
How we address the issue of lack of personalization
Drawing on our combined strengths in personality science, computer science, and medical education, PerTRAIN applies theory-guided LLMs that simulate realistic patient personalities within medical encounters. These models capture differences in symptom expression (vivid vs. understated descriptions of pain or discomfort) and communication style (friendly, anxious, reserved, or direct) based on scientifically established personality dimensions. We use a hybrid development strategy: Personality theory defines clear behavioral markers, while data-driven training refines them using recordings and transcripts from real interactions and actor-based simulations. Models are iteratively prompted, fine-tuned or audited to accurately mimic the intended personalities.
How the personalization trainings work
Simulated AI patients are embedded in progressively complex simulation environments:
- Chat-based applications
- Simulated phone/video consults
- Immersive AR/VR scenarios
- Noticing personality cues in language and behavior
- Adapting communication to the individual and situation
- Reflecting on decisions to consolidate learning
How we ensure credibility
We apply methods we’ve used in prior translational projects: expert ratings, technical assessments, and comparisons between model behavior and real behavior. Additional evaluations (reaction times, physiological indicators, self-reports) help confirm immersion and believable personality portrayals. Findings feed a continuous improvement loop between our psychology, computer science, and medical education teams.
Path to adoption
PerTRAIN is being prepared for broader use across medical education and professional development, beginning with focused collaborations and pilots. As evidence accumulates, we plan to expand access through established educational channels and continuing-education frameworks, with the aim of informing future standards for training personalized clinical interactions in the health care system.
NewsFunding & Support
Our Sponsors
This research is generously supported by:
We are grateful for the support that makes this research possible.