Rough Classification of HSV Patients

Abstract

An information system containing 122 patients with duodenal ulcer treated by highly selective vagotomy (HSV) is analyzed with the concept of rough sets. Twelve attributes are used to describe the patiens: 11 attributes concern anamnesis and preoperative gastric secretion and 12th attribute defines classification of patients according to long term results of the operation in the Visick grading. Using the methodology based on the rough sets theory, the information system is reduced so as to get a minimum subset of attributes ensuring an acceptable quality of the classification. A “model” of patients in each class is constructed upon the analysis of values adopted by attributes from this subset. Then, the reduced information system is identified with a decision table, assuming that the attributes in the minimum subset are condition attributes and that the result of treatment is a decision attribute. From this table, a decision algorithm is derived, composed of 44 decision rules. The algorithm and the models are helpful in decision making concerning the treatment of new duodenal ulcer patients by HSV.