Research on ECG Signal Classification Based on Hybrid Residual Network

Arrhythmia detection in electrocardiogram (ECG) signals is essential for monitoring cardiovascular health. Current automated arrhythmia classification methods frequently encounter difficulties in detecting multiple cardiac abnormalities, particularly when dealing with imbalanced datasets. This paper proposes a novel deep learning approach for the detection and classification of arrhythmias in ECG signals using a Hybrid Residual Network (Hybrid ResNet). Our method employs a Hybrid Residual Network architecture that integrates standard convolution, depthwise separable convolution, and residual connections to enhance the feature extraction efficiency and classification accuracy. To guarantee superior input signals, we preprocess the ECG signals by removing baseline drift with a high-pass Butterworth filter, denoising via discrete wavelet transform, and segmenting heartbeat cycles through R-peak detection. Additionally, we rectify the class imbalance in the MIT-BIH Arrhythmia Database by applying the Synthetic Minority Oversampling Technique (SMOTE), therefore enhancing the model’s ability to detect infrequent arrhythmia types. The suggested system achieves a classification accuracy of 99.09% on the MIT-BIH dataset, surpassing conventional convolutional neural networks and other state-of-the-art methodologies. Compared to existing approaches, our strategy exhibits superior effectiveness and robustness in managing diverse irregular heartbeats and arrhythmias.