ECG beat classification by using autocorrelation and RR intervals

The classification of Electrocardiogram (ECG) is vital for medical examination or monitoring of critical ill patients due to changes in the normal rhythm of a human heart, when associated with heart attack, may lead to mortality. Consequently, feature extraction and automatically identify arrhythmias turn out to be much more important. In this study, RR interval series and autocorrelation are used to extract features. Subsequently, we applied least square support vector machine(LSSVM) classifier to discriminate the detected features into normal or premature ventricular contraction (PVC). The results obtained were: the total sensitivity is 99.64%, 97.17% in positive predictive value and 3.17% in error rate.