Xecgarch: A Multi-scale Convolutional Neural Network Cnn For Accurate And Interpretable Atrial Fibrillation Detection In Ecg Analysis

A team of researchers has developed an interpretable deep learning architecture, xECGArch, for accurate and trustworthy ECG analysis. This approach utilizes deep Taylor decomposition to highlight key features and bridge the gap between clinical needs and automated analysis. The architecture achieved a 95.43% F1 score on unseen data and was tested on four extensive ECG databases.