Complementary, Not Competing: The Practical Value of AI-Powered Virtual Patient Simulation Alongside Manikin and Human Simulation in Health Professions Education

Abstract

Simulation is central to modern health professions education, and high-fidelity manikins and human simulation, including standardized patients, have a strong and well-earned evidence base for developing psychomotor skills, teamwork, and clinical judgment. These modalities are resource intensive, however, requiring dedicated labs, equipment, and substantial faculty time, which constrains how much practice each learner can receive. Screen-based virtual patient simulation, and in particular the new generation powered by large language models, offers a complementary approach that is highly scalable, lower in marginal cost, available on demand, and able to capture detailed interaction data for assessment and program evaluation. This article gives full credit to manikin and human simulation while arguing that AI-powered virtual patients add important practical value and, for cognitive and communication outcomes such as knowledge, clinical reasoning, and history taking, can achieve a comparable learning yield. Drawing on meta-analytic evidence that instructional design rather than physical fidelity is the primary driver of learning, randomized trials of virtual and AI-simulated patients, and recent evaluations of large language models in medicine, I propose a blended model in which each modality is used for what it does best.