Dispersed decision-making system with selected fusion methods from the measurement level-Case study with medical data

Abstract

In the paper issues related to the use of dispersed knowledge in medicine are discussed. The main aim of the article is to investigate the efficiency of inference of seven selected fusion methods in a dispersed decision-making system. The dispersed system was proposed by the author in previous papers. The examined fusion methods-the maximum rule, the minimum rule, the median rule, the sum rule, the probabilistic product method, the method that is based on the theory of evidence and the method that is based on decision templates-are well known from the literature. In the paper two medical data sets from the UCI repository were used. Based on the obtained results it was concluded that for one data set the maximum rule generates the best results, and for other data set better methods are the sum rule and the median rule.