Detection of cardiac arrhythmia using convolutional neural network

Abstract

Cardiac arrhythmia (abnormal heart rhythm), which may even life threatening sometimes. An automatic diagnosis system is required to identify the cardiac arrhythmia at early stages for immediate precaution and treatment. With the use of novel machine learning algorithms, we classified different types of cardiac diseases. Detection of heart arrhythmia requires preprocessing, feature extraction and classification steps. Feature extraction step plays a major role in accurate detection of arrhythmia, as feature extraction methods provide us a way of reducing computation time, increasing prediction performance, and provides a detailed understanding of the disease. Discrete wavelet transform (DWT) is used as feature extraction technique, and extracted features are classified using SVM and KNN. We applied the feature extraction and classification techniques on the standard MIT-BIH datasets of (myocardial infarction) to demonstrate the applicability of feature extraction techniques for the detection of abnormal heart rhythm.