ECG Arrhythmia Detection with Machine Learning Algorithms

Abstract

Arrhythmia is one of the major causes of deaths across the globe. Almost 17.9 million deaths are caused due to cardiovascular diseases. This study has been conducted to classify heartbeats into two classes, the one with regular heartbeat and the other having irregular heartbeat. The dataset that is used here has been collected from California University at Irvine Machine Learning Data Repository. First of all, the dataset is pre-processed in which the data normalization is performed and the missing values are removed. Following the previous step, feature selection is performed by Principal Component Analysis (PCA). Then, 8 classifiers are applied on the various data splits. Finally, Accuracy, Sensitivity, and Specificity are calculated. The maximum accuracy of 89.74% is obtained using SVM and Naïve Bayes after applying feature selection method on 90-10 data split.