ECG Signal Denoising and Feature Learning for Accurate Arrhythmia Classification

Proactive diagnosis of diseases with artificial intel-ligence and associated technologies has emerged as a potential area in recent years. Cardiovascular diseases require continuous monitoring and are a key focus area in medical care. Arrhythmia, a cardiovascular disease, is typically diagnosed by physicians using Electrocardiography, which monitors the electrical activity and rhythm of the heart. Neural networks have been used extensively recently to identify anomalies in electrocardiogram (ECG) data. Previous research has shown that using a denoised ECG signal rather than raw data significantly enhances the likelihood of detecting arrhythmia. This paper evaluates four filters applied to ECG signals to improve classification accuracy. Further, we have developed a custom deep-learning technique for ECG data analysis. We conducted extensive experiments using these four filters and the proposed deep neural network technique. The comparative analysis demonstrates that the pro-posed method outperforms existing methods across multiple performance metrics.