Virtual Medical Board: A Distributed Bayesian Agent Based Approach

Abstract

Distributed Decision Making has become of increasing importance to get solution of different real life problems, where decision makers are in geographically dispersed locations. Application of agent and multi agent system in this Distributed Decision Support System is an evolving paradigm. One of such real life problem is medical diagnosis. For critical medical diagnosis, medical board is formed which is a coordinative discussion mechanism between a group of expert physicians to diagnose a patient. But always forming a medical board with a group of expert physicians may not be possible due to lack of infrastructure, availability, time etc. In that situation the role of multi agent based distributed decision making can comes into play. In this paper we develop a Virtual Medical Board System in which a number of software agents (expert agents) act as a group of expert physician with knowledge base(KB), reasoning capability. They coordinatively discuss with each other to diagnose a patieh each other to diagnose a patient. We represent the discussion module of the system in the form of Bayesian Network of Bayesian Agent (BNBA). In BNBA each BA is the expert software agent whose Knowledge Base (KB) is represented in the form of Bayesian Network (BN). Also the BDI (Belief, Desire, Intention) model of each BA is represented in this paper.