Enhancing accuracy of arrhythmia classification by combining logical and machine learning techniques

This paper is a contribution to the Physionet/Computing in Cardiology Challenge 2015. The aim is to reduce the occurrence of false alarms in the ICU during the detection of asystole, extreme bradycardia, extreme tachycardia, ventricular fibrillation and ventricular tachycardia. Robust classification of each arrhythmia is achieved using a combination of logical and SVM-based machine learning techniques. Information from electrocardiogram and photoplethysmogram signals, sampled at 250Hz, is used for logical analysis and to form the feature set. This information includes time-domain and frequency-domain data such as R-R intervals, power spectrum density, autocorrelation plots and standard deviation values. Pan-Tompkins algorithm is applied to ECG signals for QRS complex detection.