AI-assisted Simulated Patient for Clinical Communication Skills Training

The ‘AI-assisted Simulated Patient’ project set out to develop a platform that allows healthcare students to practise clinical communication through realistic, AI-driven patient simulations. The system provides automated feedback and supports replay, repetition and reflection - allowing learners to engage with scenarios at their own pace in a psychologically safe environment.

Effective and essential communication skills are often difficult to teach to healthcare students. Textbooks can only go so far, and traditional methods such as live observed simulations can be stressful, especially for students with neurodevelopmental conditions, mental health challenges or learning differences. Remote learners also face barriers to accessing regular in-person simulation-based training.

With technical support from Edward Fauchon-Jones, Senior Research Software Engineer, and Dominik Lukes, Lead Business Technologist, at the AI Competency Centre, the project team built and tested a prototype that delivered lifelike patient responses, including natural inflections.

Students responded positively, embracing the opportunity to experiment with communication techniques in a low-pressure setting. However, the team also encountered several challenges. Calibrating the AI to respond appropriately - neither oversharing unsolicited information nor being too evasive - was a delicate balance. Simulating complex patient behaviours, such as anger or resistance, proved difficult, as language models tend to default to cooperative dialogue.

Key lessons emerged around the importance of clear instructions, thoughtful scenario design, and integrating pedagogically meaningful feedback. With this foundation, AI tools such as this one can significantly enhance experiential learning, improve accessibility, and boost engagement in clinical education.