Diagnosis of cardiac arrhythmia using fuzzy immune approach

Abstract

Classification is an important data mining task in biomedicine. For easy comprehensibility, rules are preferrable to another functions in the analysis of biomedical data. The aim of this work is to use a new fuzzy immune rule-based classification system for a medical diagnosis of a cardiovascular disease. In this study, fuzzy immune approach (FIA), which can be improved by ours, is a new method and firstly, it is applied to ECG dataset. The performance of the proposed approach, in terms of classification accuracy, ROC curves, and area under the ROC curve (AUC) was compared with traditional classifier schemes: C4.5, Naïve Bayes, KStar, Meta END, and ANN. The classification accuracies and AUC statistics of FIA for the data sets used are the highest among the classifiers reported on the UCI website and other classifiers used for related problems and tested by cross validation. © 2011 Springer-Verlag.