Optimizing Electrocardiogram Denoising for Enhanced Cardiovascular Disease Detection: A Metaheuristic Approach

Cardiovascular disease (CVD) is the leading cause of death worldwide, accounting for more deaths than any other known cause. Hence, early detection followed by timely treatment of these diseases is crucial to preventing premature deaths. In this scenario, the electrocardiogram (ECG) emerges as a key diagnostic tool, providing critical insight into the heart’s electrical activity and allowing early identification of potentially lethal conditions such as arrhythmias and heart attacks. The automated analysis of ECGs represents a potential tool for the timely detection of different heart conditions. Nevertheless, noise is always present due to the signal acquisition process, and the degree of removal highly impacts the ECG classification accuracy. This paper presents an approach to determining the best ECG degree of noise removal effectively. It comprises the iterative analysis of the wavelet-based denoising method and the Extreme Gradient Boosting (XGBoost) classification, whose best noise removal parameter configuration is obtained through an optimization based on metaheuristic algorithms (MAs). Different MAs are tested to evaluate their performance in classification accuracy enhancement. This proposal is trained and tested on the MIT BIH public ECG dataset to demonstrate its effectiveness across different signal acquisitions. This method is intended to be a preprocessing stage to improve the accuracy of predictive models based on neural networks and the future development of more robust ECG classifier systems, which will improve the detection of CVD.