Paper Number
ECIS2026-2331
Paper Type
CRP
Abstract
Effective doctor-patient communication is a critical determinant of patient’s health, satisfaction, and safety, because it reduces medical errors and is as highly valued by patients as clinical competency. However, traditional communication training with human actors is resource-intensive and often inaccessible, leaving a gap in medical education. Generative AI offers a transformative alternative by enabling high-quality communication simulations. This paper develops and evaluates design principles for generative AI-based virtual patients to deliver realistic, scalable, and personalized training. We evaluated our design knowledge through an instantiation of the principles in a generative AI-based prototype that provides access to over 500 realistic patient scenarios. In a study with 39 medical students and practicing doctors, we evaluated the realism, usefulness, and feedback quality. This work provides prescriptive guidance for integrating disruptive technologies into medical education, supporting the digital transformation of healthcare training and equipping future physicians with essential communication skills for improved patient care.
Recommended Citation
Laufer, Jan; Banh, Leonardo; Strauß, Volker; and Strobel, Gero, "Grey’S A(I)Natomy: Design Principles For Generative Virtual Patients" (2026). ECIS 2026 Proceedings. 15.
https://aisel.aisnet.org/ecis2026/hit/hit/15
Grey’S A(I)Natomy: Design Principles For Generative Virtual Patients
Effective doctor-patient communication is a critical determinant of patient’s health, satisfaction, and safety, because it reduces medical errors and is as highly valued by patients as clinical competency. However, traditional communication training with human actors is resource-intensive and often inaccessible, leaving a gap in medical education. Generative AI offers a transformative alternative by enabling high-quality communication simulations. This paper develops and evaluates design principles for generative AI-based virtual patients to deliver realistic, scalable, and personalized training. We evaluated our design knowledge through an instantiation of the principles in a generative AI-based prototype that provides access to over 500 realistic patient scenarios. In a study with 39 medical students and practicing doctors, we evaluated the realism, usefulness, and feedback quality. This work provides prescriptive guidance for integrating disruptive technologies into medical education, supporting the digital transformation of healthcare training and equipping future physicians with essential communication skills for improved patient care.