Development of Virtual Patient Simulations for Medical Education

May 10, 2011 3:20 - 5:00pm

Douglas Danforth Ph.D., Obstetrics and Gynecology
Mike Procter, B.S., Athabasca, Alberta
Robert Heller, Ph.D., Athabasca, Alberta
Eloise Pasteur, Ph.D., York, United Kingdom
Richard Chen, M.D., Columbus, OH
Mary Johnson, Ph.D., Florida State University Chad Jackson, M.S., RRT, CHT, American College of Chest Physicians

The principal goal of this project is the development of virtual patient simulations for medical education. In order to simulate real patients with greatest fidelity, virtual patients in Second Life (SL) are controlled by artificial intelligence. This allows students to engage in a natural language conversation with the patient to obtain relevant patient history, symptoms, etc, and then to develop differential diagnoses and treatments appropriate for the simulated condition of the patients. Our goal is to provide students with undifferentiated patients to sharpen their diagnostic skills as well as allowing students to rehearse professional behaviors in a risk-free environment, providing opportunities for skills practice and feedback prior to real-world patient encounters. The system consists of; 1) a web interface that allows subject experts to build and maintain the questions and case datasets without having to understand or interact with the underlying code, 2) a case repository that contains the content, or knowledge, that the virtual patient needs to perform its task, decoupled from AIML or any specific conversational engine. 3) an AIML generator which is a Java application that converts XML case files to AIML for the conversational agent. This content AIML is combined with common AIML files which handle generic conversation-related behaviors and avatar control. 4) an AIML sever which is an open source program – (Program D) with an HTTP server component. SL scripts access this service via typical HTTP request functions, and 5) Second Life scripts which provide an interface between SL users and AIML server, avatar animation and other interaction with avatars and in world objects.