Combination of neural networks and fuzzy sets as a basis for medical expert systems

Presents a technique for building expert systems which combines the fuzzy-set approach with artificial neural network structures. This technique can effectively deal with two types of medical knowledge, a nonfuzzy one and a fuzzy one, which usually contribute to the process of medical diagnosis. Nonfuzzy numerical data are obtained from medical tests. Fuzzy linguistic rules describing the diagnosis process are provided by a human expert. The proposed method has been successfully applied in veterinary medicine as a support system in the diagnosis of canine liver diseases