Classification analysis is widely a

Classification analysis is widely adopted for healthcareapplications to support medical diagnostic decisions,improving quality of patient care, etc. A subset dataset of theextensive amounts of data stored in medical databases isselected for training. If the training dataset contains irrelevantfeatures, classification analysis may produce less accurate andless understandable results. Feature subset selection is one ofdata preprocessing step, which is of immense importance inthe field of data mining. This paper proposes the filter andwrapper approaches with Particle Swarm Optimization (PSO)as a feature selection methods for medical data. Theperformance of the proposed methods is compared withanother feature selection algorithm based on Genetic approach.The two algorithms are applied to three medical data sets Theresults show that the feature subset recognized by theproposed PSO when given as input to five classifiers, namelydecision tree, Naïve Bayes, Bayesian, Radial basis functionand k-nearest neighbor classifiers showed enhancedclassificati