ECG-COVID: An end-to-end deep model based on electrocardiogram for COVID-19 detection

Area of Science:

• Cardiology
• Artificial Intelligence
• Infectious Disease Diagnostics

Background:

• Accurate and early detection of COVID-19 is crucial for controlling its spread and lifting restrictions.
• Current diagnostic methods for COVID-19 have limitations.
• Electrocardiogram (ECG) signals offer a non-invasive and accessible approach for potential COVID-19 detection.

Purpose of the Study:

• To propose a novel deep learning model, ECG-COVID, for accurate COVID-19 detection using ECG signals.
• To evaluate the efficacy of an end-to-end deep learning approach for COVID-19 screening.
• To provide a supplementary diagnostic tool for managing COVID-19 patients.

Main Methods:

• Developed an end-to-end deep learning model named ECG-COVID.
• Employed multiple deep Convolutional Neural Networks (CNNs) on a dataset of 1109 ECG images.
• Selected the most efficient CNN model for final evaluation without additional processing stages.

Main Results:

• The ECG-COVID model achieved an average accuracy of 98.81%.
• The model demonstrated high performance with Precision, Sensitivity, and F1-score all at 98.8%.
• The end-to-end approach directly processed ECG images for COVID-19 detection.

Conclusions:

• The proposed ECG-COVID model shows significant potential for accurate and efficient COVID-19 detection.
• This non-invasive ECG-based method can aid in rapid screening, especially in high-demand hospital settings.
• The study highlights the utility of deep learning in analyzing physiological signals for disease diagnosis.