We present an overview of the Virtual Patient project at the University of Maryland, which is developing a cognitive model of humans experiencing various states of health and disease to be used in interactive simulations for physician training. Overview This Virtual Patient1 project is devoted to creating a cognitive, knowledge-based model of a virtual patient (VP) that undergoes both normal and pathological physiological processes. VPs are ontological objects, specifically, subclasses of VIRTUAL-HUMAN that have various diseases and disorders. Like all VIRTUALHUMANs, their large inventory of property-value pairs changes in response to ontological events, including internal and external stimuli. All VPs inherit the lion's share of physiology from VIRTUAL-HUMAN, meaning that GERD-PATIENT and HEART-DISEASE-PATIENT (as ontological concepts, not instances) differ only with respect to the disease-specific changes that affect certain of their property values over time. A cornerstone of ...