Current Classification Algorithms for Biomedical Applications

In this study we report the advances in supervised learning methods that have been devised to analyze medical data sets. As mining of data sets produced by medical equipments is becoming an increasingly challenging task, due to the size of the databases and the gradient of their update, new methods need to provide classification models that can handle the complexity of the problems. We start describing standard methods and we show how kernel methods, incremental learning algorithms and feature reduction techniques, applied to standard classification techniques, can be successfully used to discriminate biological and medical data sets. Among existing methods, we describe those that have their foundations in the statistical learning theory and have been successfully applied to the field. We provide numerical experiments based on publicly available data sets, and discuss results in terms of classification accuracy. Finally, we draw conclusions and outline future research directions.