Classification of Tachycardia, Bradycardia and Premature Ventricular Arrhythmias using Support Vector Machine

An electrocardiogram (ECG) signal is one of the most important non-invasive tools for diagnosing cardiac arrhythmias. This article is about the automatic classification of arrhythmias and premature ventricular arrhythmias (PVC). QRS detection is performed using the Pan Tompkins algorithm. Heart beat is identified by three consecutive RR characteristics related to the current heartbeat, the previous heartbeat, and the next heartbeat. Classification is performed based on the percentage of the desired heart beat in order to identify patients with significant risk factors. Multi-class support vector machine (SVM) with one-to-one (OAO) approach is used to classify type of arrhythmias. The measured data were extracted from the MIT-BIH database. The proposed method has an accuracy of 97.92 and 95.83 for heart beat arrhythmia, PVC arrhythmia detection and a total accuracy of 96.87, respectively