ECG Monitoring System Based on Multi-Scale Convolutional Neural Network

This paper introduces the Multi-Scale SEEnhanced Convolutional Network with Atypical Pre-Activation (MSE-CNet-AP), a novel model for a multi-lead ECG signal recognition. The proposed system applies a multi-scale network with attention mechanisms to enhance spatial feature extraction from ECG signals, which are then used for classification. The proposed system uses a single-board computer as a hardware part, and facilitates real-time ECG data acquisition and classification, achieving a 96.40% identification accuracy, highlighting its clinical potential.