Gabor Wavelet Scattering Network and KNN-Based Arrhythmia Classification Model

This study presents an automated method for enhancing the precision and speed of electrocardiogram (ECG) interpretation, aiming to assist in the early identification of cardiovascular diseases (CVDs). The proposed model leverages the Gabor Wavelet Scattering Transform (GWST) in conjunction with K-nearest neighbors (KNN) for detecting congestive heart failure (CHF) and arrhythmia rhythm (ARR) within ECG data. The research employs datasets from the MIT-BIH Arrhythmia database, MIT-BIH Normal Sinus Rhythm (NSR), and Beth Israel Deaconess Medical Center (BIDMC). To assess the model's performance, a five-fold cross-validation approach is employed. The validation results demonstrate an impressive 100% accuracy and F1 score. Furthermore, when independently tested on the MIT-BIH and BIDMC datasets, the model exhibits outstanding accuracy at 99.94% and an F1 Score of 99.93%. These outcomes underscore the remarkable accuracy and F1 scores achieved by the proposed model in both validation and independent testing across the MIT-BIH and BIDMC datasets.