Computational Intelligence for Cardiac Arrhythmia Classification

This paper presents a comparative study of automatic classification of different types of heart beat arrhythmias. The heart beats are classified into normal, premature ventricular contraction, atrial premature, right bundle branch block and left bundle branch block classes. Different classifiers are used in this work, namely support vector machine, multilayer perceptron neural networks, and TreeBoost. We carried out several experiments using the MITBIH arrhythmia database and obtained promising results. The computed average accuracy, sensitivity, and specificity are 98.89%, 90.63%, and 98.71%, respectively. Results have demonstrated that TreeBoost and support vector machine have an edge over multilayer perceptron neural networks for arrhythmia classification. Keywords-component; Arrhythmia Classification; support vector machine; multilayer perceptron; TreeBoost; ECG signals.