Classification of Electrocardiogram Cardiac Arrhythmia Signals Using Genetic Algorithm - Support Vector Machines

This research study has been focused on exploring the novel approach for classifying the arrhythmia disease of cardiac muscle. The proposed methodology determines the combination of Genetic Algorithm and Support Vector Machines techniques. Initially, the feature extraction of twenty-four features has been made from the ECG waveform. The acquisition of these features has been made by partial automatic extraction from the amplitude (voltage)-time parameters of P peak, Q peak, R peak, S peak and T peak feature sets of an ECG waveform. Genetic Algorithm is utilized for improving the performance of generalization in the Support Vector Machine classifier. In order to proceed with this task, the optimization of Support Vector Machine classifier is done by performing the search of the parameter with the best value which makes tuning of its discriminate function and seeking for the best feature subset with which in response does the optimization of the fitness function associated with the classification results. Certain simulations have been carried out with the help of MATLAB software with which the results over the experiments determines the demonstration that the proposed methodology does the best classification of ECG waveforms in detecting the cardiac arrhythmias. The recorded ECG dataset has been acquired from MIT-BIH arrhythmia database and 4 different sorts of arrhythmias has been considered for performing the classification task and it is obtained with the results of 97.45% of accuracy, 95.2% of sensitivity, 95.4% of specificity and 96.23% of positive predictivity.