Expert system for pathological bacterial testing and determination of corresponding antimicrobial therapy

Abstract

This is an attempt to develop an expert system for computer-guided of serial pathological bacterial testing to facilitate determination of antimicrobial therapy. If deem successful, the system could indirectly help reduce the cost of pathological testing and increase the efficiency of assignment of antimicrobial therapy to patients. Six common pathological tests, 19 infectious bacteria, and 16 recommended antimicrobial therapies were screened, identified, and selected. Decision table and decision tree are used in developing the knowledge base. Bayesian reasoning is employs for uncertainty management. Four user interface testers and one domain expert reviewer tested the usability of the system. The system has good potential but required further rigorous knowledge acquisition techniques, more breadth and depth in knowledge base contents, and better help facilities.