Fuzzy neural networks versus alternative approaches in medical decision support

Abstract

One of two goals of this paper is to briefly present a methodology for medical decision support systems design which is able to utilize two main types of medical knowledge usually contributing to medical diagnosis: a qualitative one (linguistic rules provided by human experts) and a quantitative one (numerical data obtained from medical tests). This methodology is based on fuzzy neural networks and has been successfully applied to the design of a support system for the treatment of duodenal ulcer with the use of highly selective vagotomy. The second goal of the paper is to carry out a broad comparative analysis of the proposed methodology with several alternative approaches (rough sets, discriminant analysis, location model, probabilistic inductive learning).