Maryland Virtual Patient: A Multi-Agent Architecture for Creating, Diagnosing and Treating Virtual Patients

 

Maryland Virtual Patient: A Multi-Agent Architecture for Creating, Diagnosing and Treating Virtual Patients

Sage Labs 4101

October 3, 2007 12:00 PM - 1:30 PM

Marjorie McShane
Institute for Language and Information Technologies and
Department of Computer Science and Electrical Engineering
University of Maryland Baltimore County

Abstract: This talk will provide a demonstration and overview of the Maryland Virtual Patient (MVP) environment, which provides interactive, open-ended simulations of virtual patients for the training and assessment of medical personnel. The environment is implemented as an agent network that includes artificial agents with cognitive capabilities (the cognitive side of the virtual patient, the mentor and various medical personnel), artificial agents with deterministically modeled physical reality (the physiological side of the virtual patient, disease processes, environmental events) and artificial agents that implement the simulation (the scheduler, the executor, etc.). The knowledge in the system is modeled using the ontological substrate of the OntoSem environment, which supports not only simulation and decision-making but also the natural language processing requirements of interagent communication and automatic mentoring.

Paper: Maryland Virtual Patient: A Multi-Agent Architecture for Creating, Diagnosing and Treating Virtual Patients

Source: paper

Add to calendar
Share|