Early Detection of Cardiac Abnormalities in COVID-19 Patients by Autoencoder Based Analysis of ECG Patterns

Abstract

Heart diseases have been greatly affected by COVID-19 and many people have suffered from heart problems. Since COVID-19 affects the cardiovascular system and can cause serious complications, it is important to diagnose cardiovascular abnormalities in patients as soon as possible. This research study presents a method to identify abnormalities in electrocardiogram imaging in COVID-19 patients using a deep learning model-based autoencoder. Autoencoders learn compressed representations of electrocardiogram patterns, ready for unsupervised feature extraction. The model distinguishes abnormal patterns from a normal baseline by reconstructing electrocardiogram (ECG) images and using new techniques to identify abnormalities. While testing and training on electrocardiogram images from COVID-19 patients, the model was shown to be effective in diagnosing heart disease with high accuracy (91.11%). This technology can facilitate immediate diagnosis of electrocardiogram abnormality in COVID-19 patients and help doctors conduct risk assessment and early diagnosis.