Efficient Cardiac Arrhythmia Detection Using Machine Learning Algorithms

Abstract

The most common type of chronic and life-threatening disease is cardiovascular disease (CVD). For the early prediction of arrhythmia, electrocardiogram (ECG) is recorded from the patients, non-invasively using surface electrode. In this approach, Empirical Mode Decomposition (EMD) is performed for noise removal followed by Pan Tompkins algorithm for feature extraction. To reduce the amount of signal characteristics and computation time, Principal Component Analysis (PCA) is utilized. Finally, two classifiers, The Support Vector Machine (SVM) and the Naive Bayes (NB) classifier is used to determine the cardiac abnormality from the ECG signal. The comparison is made between the two classifiers and their accuracy will be analysed. We obtained 89% accuracy for SVM and 99% for NB classifier. Lakhs of samples will be available in the Physionet. The amplitude of the signal is 0.1 Mv and time period (T) is 10ms and the frequency of 100Hz. The Confusion Matrix can then be used to assess how well an ECG signal is performing. A MATLAB program is used which has the capacity to observe the ECG bio-signal on a computer.