Classification analysis is widely adopted for healthcare applications to support medical diagnostic decisions, improving quality of patient care, etc. A subset dataset of the extensive amounts of data stored in medical databases is selected for training. If the training dataset contains irrelevant features, classification analysis may produce less accurate and less understandable results. Feature subset selection is one of data preprocessing step, which is of immense importance in the field of data mining. This paper proposes the filter and wrapper approaches with Particle Swarm Optimization (PSO) as a feature selection methods for medical data. The performance of the proposed methods is compared with another feature selection algorithm based on Genetic approach. The two algorithms are applied to three medical data sets The results show that the feature subset recognized by the proposed PSO when given as input to five classifiers, namely decision tree, Naïve Bayes, Bayesian, Radial basis function and k-nearest neighbor classifiers showed enhanced classification accuracy over all given types of classification methods.

Feature selection Particle Swarm Optimization medical datasets Decision tree Naïve Bayes Bayesian Classifier Radial Basis Function K-Nearest Neighbor.