Ecg Signal Classification Using Ensemble Decision Tree

The electrocardiogram (ECG) is a non-invasive method to measure and record the electrical activity of the heart. ECG signal analysis has an important role on the diagnosis of heart diseases especially, abnormal or irregular heartbeats, namely arrhythmia. There are three basic waves; P, QRS and T in healthy EGC signal. The detection of these waves and time domain morphological properties represent the information about arrhythmia. Time intervals between waves or duration of a wave such as RR interval (RR) and QRS length are successful and well-studied methods of detecting arrhythmia. In addition, form factor (FF) is another technique to represent ECG waveform complexity in a scalar value. In this paper, arrhythmia beat classification using ensemble decision tree is studied. Bootstrap aggregating (bagging) decision tree is used as a type of ensemble learning. ECG signals from 22 patients including five arrhythmia beats and normal beats are obtained from MIT-BIH arrhythmia database. After the filtering process, 56569 ECG beats are collected and feature are extracted based on morphological properties including RR, FF, RR and FF ratio to previous values (RRR, FFR), RR and FF differences from mean values (RRM, FFM). 25% of 56569 beats is used as test data for bagged decision tree and the rest for training. The performance measures of bagged decision tree with varying 75 learners and single decision tree are evaluated to compare the effect of bagging decision tree on ECG beat classification. While bagged decision tree gives accuracy of 99.34%, decision tree yields 98.30% accuracy. Finally, we observe that the bagged decision tree for ECG arrhythmia beat classification can be successfully applied to increase the accuracy of ECG arrhythmia detection.