A Bayesian expert system for the analysis of an adverse drug reaction

摘要

We describe a prototype probabilistic expert system which has been developed for analysing an adverse drug reaction problem, specifically, drug-induced pseudomembranous colitis. The system incorporates two simple Bayesian networks to model the main influences suspected to play a leading role in the adverse drug reaction. It uses the Lauritzen-Spiegelhalter algorithm for manipulating probability distributions in Bayesian networks.