Intelligent medical patient interviewer

摘要

This article is about our current efforts in medical expert systems. The domain is that of an initial patient interview to elicit information required to build and maintain a patient record for a medical practice database. The challenge is to replace the manual process of having the patient fill out a lengthy questionnarire, then transferring the data to a paper or electronic record. Interestingly enough, no existing systems address this domain. Our approach is a segmented knowledge base with screening questions and follow-up question sets. Screening questions fire rules which trigger follow-up question sets. Implementation is planned in three stages. We are in the first stage, creating a prototype in an expert system shell. Effort is split between knowledge base and user interface development, since the patient is the intended user, with liittle support from the medical staff. The second stage will expand the program, incorporating filters for compatibility with existing database systems. The third stage will develop a speech recognition system for input. Currently we have the preliminary screens for the user interface and an approach to segmentation of the knowledge base, an estimated 500-1000 rules.