ECG Beat Classification with Fractional Order Differentiator and Machine Learning Techniques

Electrocardiogram (ECG) is essential for assessing heart function, but manual analysis is time-consuming and error-prone. Automated ECG analysis can improve early detection of cardiovascular diseases by accurately identifying abnormal beats despite signal irregularity and non-stationarity. In this work, a novel approach for accurate ECG beat classification was proposed, integrating a sequential approach with a fractional order differentiator, dual-tree complex wavelet transform (DTCWT) features, and machine learning (ML) classifiers. This methodology involves R-peak detection using a fractional order differentiator, feature extraction with DTCWT, and classification using various ML models. Evaluated on the MIT-BIH Arrhythmia Database, this approach demonstrates superior performance, with the Random Forest classifier achieving an accuracy of 96.82%, sensitivity of 96.83%, specificity of 97.02%, PPV of 96.89%, and an F1 score of 96.85%. These results underscore the effectiveness of this approach in improving the accuracy of ECG beat classification, contributing to better clinical outcomes in heart disease diagnosis.