Classification and Prediction of Cardiovascular Autonomic Neuropathy Severity Using Fuzzy Logic-Based Expert System

The present article aims at designing an expert system for diagnosis of Cardiovascular Autonomic Neuropathy (CAN) severity. A data set was synthesized representing the significant characteristics of some study subjects with varying CAN severity and considered in this study. The three-step design procedure consists of dimension reduction of data set using Sammon's non-linear mapping (NLM) followed by entropy-based fuzzy C-means clustering of the mapped data set and finally, designing the expert system. The design parameters of the expert system were optimized using a Particle Swarm Optimization (PSO) off-line, which could later be deployed for prediction of the severity of CAN on-line. The characteristics with notable influence on CAN were utilized by the expert system as inputs and the output in the form of the grade/degree of CAN severity was obtained.