An Intelligent ECG-Based Tool for Diagnosing COVID-19 via Ensemble Deep Learning Techniques

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Medicine
• Cardiology

Background:

• Accurate and rapid diagnosis of COVID-19 is crucial for public health and healthcare system management.
• Existing COVID-19 diagnostic tools have limitations, necessitating the exploration of novel methods.
• Cardiovascular alterations observed in COVID-19 patients suggest the potential utility of electrocardiogram (ECG) data for diagnosis.

Purpose of the Study:

• To introduce a novel automated diagnostic tool for COVID-19 detection using ECG data.
• To leverage deep learning (DL) models and machine learning classifiers for enhanced diagnostic accuracy.
• To evaluate the efficacy of ECG-based diagnosis for differentiating COVID-19 from normal and other cardiac conditions.

Main Methods:

• Utilized ten deep learning (DL) models of diverse architectures to extract significant features from ECG data.
• Implemented a hybrid feature selection method combining chi-square test and sequential search.
• Employed machine learning classifiers for binary (COVID-19 vs. normal) and multiclass (COVID-19 vs. normal vs. other cardiac complications) classification.

Main Results:

• The automated tool achieved 98.2% accuracy in binary classification (normal vs. COVID-19).
• The tool reached 91.6% accuracy in multiclass classification (differentiating COVID-19 from normal and other cardiac conditions).
• Demonstrated the potential of ECG data as a viable alternative diagnostic tool for COVID-19.

Conclusions:

• The proposed ECG-based automated diagnostic tool shows high performance in identifying COVID-19.
• ECG analysis, augmented by deep learning, offers a promising, non-invasive approach for COVID-19 diagnosis.
• This method can potentially alleviate burdens on healthcare systems and aid in managing the pandemic.