Deep Learning Model for ECG Arrhythmia Detection

Detecting arrhythmias is crucial for the early diagnosis and treatment of various heart conditions. This project introduces a convolutional neural network (CNN) with an attention mechanism to classify arrhythmias into two-class and five-class categories using the PTB-XL dataset. To address imbalanced data, upsampling techniques are applied. The model’s performance is evaluated using cross-validation, achieving accuracies of 84.0% for two-class and 88.8% for five-class classifications. These results, compared with four other academic studies, underscore the efficacy of the proposed method in accurately identifying arrhythmias.