Abstract

Heart Arrhythmia is a cardiovascular disease that Electrocardiogram (ECG) signal records can be used to identify and classify into categories including; normal heartbeats, supra-ventricular premature, ventricular escape, fusion of ventricular and normal, and unclassified heartbeats. However, interpreting these signals accurately remains intricate for medical professionals that examine patient ECG report to make diagnosis and to ascertain the risk level. To aid the medical professionals in making heart arrhythmia diagnosis and classification, this study leverages the prowess of machine and deep learning techniques, which have showcased their efficacy in medical domains by automating processes and enhancing result accuracy. On this pedestal, this research developed two diagnostic model using Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM). The two models were thereafter hybridized (LSTM-CNN) and evaluated. The three models were compared and it was revealed that the CNN and the hybrid LSTM-CNN emerge as notable performers, exhibiting remarkable accuracy results of up to 97% respectively compared to LSTM that showed 90%. This research signifies a significant stride toward leveraging advanced technologies for more effective and efficient cardiovascular disorder diagnosis and management.