SimPatient - High-fidelity AI patient simulation platform using hyper realistic behavioral modeling for scalable medical education

Background
The NHS Scotland faces urgent challenges in scaling medical education while maintaining high standards. As plans for workforce expansion unfold, traditional training models that rely on costly standardized patients and limited placements are increasingly unsustainable. These limitations restrict opportunities for consistent, equitable clinical skills development, particularly in underserved or rural communities. To address this, we developed SimPatient (Simulated Patients), an AI-driven virtual patient platform that offers high-fidelity, culturally diverse, and scalable training opportunities.

Technical Architecture
SimPatient integrates advanced natural language processing, real-time avatar technology, and behavioral modeling to simulate emotional and socially authentic patient interactions. Built on a fine-tuned transformer-based large language model trained on over 60 hours of real clinical consultations, SimPatient produces nuanced, lifelike patient dialogues. It supports multimodal learning through text, voice, video, and VR interfaces. The platform’s avatar engine uses HeyGen-based photorealistic visuals to generate real patients' expressions and behaviors. A behavioral state machine adjusts emotional and physiological responses based on the learner’s communication style, fostering reflective and adaptive learning.

SimPatient is a simulation with diversity, which ensures learners interact with demographically varied patients. This includes variation in age, ethnicity, regional accents, literacy levels, and social contexts—addressing algorithmic bias and supporting training in cultural competence. Patient profiles include biopsychosocial data, non-verbal cues, and realistic symptom variation based on demographic factors. These features go beyond existing simulators, offering a more holistic and inclusive educational experience.

Designed for curricular integration, SimPatient tracks key performance indicators such as communication effectiveness, diagnostic accuracy, and empathy. Assessment is supported by validated rubrics used in clinical communication training. The initial pilot is ongoing to test with medical students (results forthcoming) interaction with and feedback for SimPatient.

Conclusion
SimPatient represents a major advancement in simulation-based medical education. It directly supports NHS goals of improving workforce efficiency, ensuring patient safety through risk-free practice, and reducing health inequalities by building students’ ability to work effectively with Scotland’s diverse population.