ADDITIONAL CRITERIA IN FUZZY 2-LEVEL PATTERN-RECOGNITION

Pattern recognition with fuzzy sets is considered with regard to medical diagnostic problems. An alternative interpretation of membership grade is suggested which expresses the degree of 'illness' of a patient with respect to a certain disease. A formal statement of a two-level fuzzy classifier is presented. The main emphasis is put on additional criteria designed to evaluate a fuzzy diagnostic decision for an unknown object. Two groups of criteria are formed - decision-making connected, and domain-dependent. The decision-making group of criteria can be viewed as related to the 'soft' consensus measure in group decisions theory. Interconnections between criteria are discussed. The notions proposed are illustrated with real medical data from the fields of aviation medicine and neonatology.