An Automated System for Arrhythmia Detection using ECG records from MITDB

Electrocardiogram (ECG) is a common non-invasive diagnostic technique used to detect cardiac disease. Several cardiac abnormalities can be uncovered by analysing the heart's electrical impulses or the combination of action impulse waveforms generated by several types of specialized cardiac tissues. Recently, cardiovascular disease has become the most prevalent disease and demands more healthcare professional to evaluate the ECG data. This has led to the widespread adoption of Machine Learning (ML)-based methods for detecting the ECG features. This research study has collected and cleaned the ECG signal data obtained from the MIT-BIH Arrhythmia database (MITB). In this work, a range of processing steps such as augmentation using SMOTE, normalization, denoising, and feature extraction using Local Binary Pattern (LBP) are used. The ML techniques employed in this study include Support Vector Machine (SVM), Linear Discriminant Analysis (LDA), and Neural Network (NN). Further, the experimental findings are utilized to establish the best model. The NN approach has been demonstrated to detect arrhythmias more accurately (92.01%) than the other three examined methods. Even when prior knowledge is inadequate, the NN approach can aid in the diagnosis of arrhythmia.