GLR-Entropy Model for ECG Arrhythmia Detection

In this paper a novel unsupervised classification method for electrocardiogram (ECG) signal classification is presented. The proposed approach classifies the input signal into normal and abnormal heartbeat patterns with a relatively high accuracy. After extracting features from the time-voltage waves in ECG signals, we utilize a computationally fast algorithm based on log likelihood strategy for change detection on selected features. We then combine the outputs based on their validation coefficient. The Algorithm could differentiate between the normal and unknown heart features. Experimental results show the accuracy of the proposed approach in terms of reliability and performance.