A Fuzzy Classifier Based on Modified Particle Swarm Optimization for Diabetes Disease Diagnosis Hamid Reza Sahebi 1 , Sara Ebrahimi 2 1 Department of Mathematics, Ashtian Branch, Islamic Azad University Ashtian, Iran sahebi@mail.aiau.ac.ir 2 Department of Mathematics, Ashtian Branch, Islamic Azad University Ashtian, Iran ebrahimi@mail.aiau.ac.ir Abstract Classification systems have been widely utilized in medical domain to explore patient’s data and extract a predictive model. This model helps physicians to improve their prognosis, diagnosis or treatment planning procedures. Diabetes disease diagnosis via proper interpretation of the diabetes data is an important classification problem. Most methods of classification either ignore feature analysis or do it in a separate phase, offline prior to the main classification task. In this paper a novel fuzzy classifier for diagnosis of diabetes disease along with feature selection is proposed. The aim of this paper is to use a modified particle swarm optimization algorithm to extract a set of fuzzy rules for diagnosis of diabetes disease. The performances of the proposed method are evaluated through classification rate, sensitivity and specificity values using 10-fold cross-validation method. The obtained classification accuracy is 85.19% which reveals that proposed method, outperforms several famous and recent methods in classification accuracy for diabetes disease diagnosis. Keywords: Diabetes disease diagnosis; Particle swarm optimization; Fuzzy classifier. 1. Introduction Diabetes is a major health problem in both industrial and developing countries, and its incidence is rising. It is a metabolic diseases characterized by high blood glucose levels, which result from body does not produce enough insulin or the body is resistant to the effects of insulin, named silent killer [1]. The body needs insulin to use sugar, fat and protein from the diet for energy. Diabetes increases the risks of developing kidney disease, blindness, nerve damage, blood vessel damage and it contributes to heart disease [2]. Early detection of diabetes is important to increase the chance of successful treatment. Such detection is often formulated as a binary classification problem[3]. Classification is a method of supervised learning producing a mapping from a feature space onto classes encountered in the classification problem. Classification problems are encountered in various domains including medicine [4], economics [5], and fault detection[6], etc. In order to improve classification performance, a large number of methods have been developed. In the classification task, the aim is assigning the patterns (case, record, or instance) to related classes, out of a set of predefined classes, based on the values of some attributes (called predictor attributes) for the patterns. There are various important categories of classification techniques including statistical techniques, neural networks, and rule based classification techniques. In recent times, many machine learning techniques have been considered to design automatic diagnosis system for diabetes. This paper specifically focuses on the use of fuzzy modeling method to detect diabetes disease which relies on discovering human comprehensible knowledge. Fuzzy approaches have become one of the well-known solutions for the classification problems. Fuzzy logic [7] improves classification and decision support systems by their allowing the use of overlapping class definitions and their powerful capabilities to handle uncertainty and vagueness. Fuzzy systems present two main advantages. First, these systems allow researchers to work with imprecise data and provide a comfortable approach to represent missing values. Second, these systems possess an interpretable rule-based structure. The performance of fuzzy classifier system depends on the “if-then” rules and their numbers that are generated from numerical data or human experiences. In the literature, several methods have been proposed to for building optimal fuzzy classifiers. In recent years Evolutionary Algorithms (EAs) have been widely used to optimize fuzzy classifiers. In the literature several EAs like Genetic Algorithm (GA), particle swarm optimization (PSO), ant colony optimization (ACO) and Artificial Bee Colony (ABC) have been proposed to produce fuzzy classification system. Rani and Deepa [8] proposed a particle swarm ACSIJ Advances in Computer Science: an International Journal, Vol. 4, Issue 3, No.15 , May 2015 ISSN : 2322-5157 www.ACSIJ.org 11 Copyright (c) 2015 Advances in Computer Science: an International Journal. All Rights Reserved.