Multi-classification method of arrhythmia based on multi-scale residual neural network and multi-channel data fusion

Area of Science:

• Cardiology
• Artificial Intelligence
• Biomedical Engineering

Background:

• Electrocardiogram (ECG) signals contain vital information for diagnosing cardiac arrhythmias.
• Accurate and automated arrhythmia detection is essential for timely medical intervention and patient monitoring.

Purpose of the Study:

• To develop an effective method for extracting ECG features and automating arrhythmia detection.
• To propose a multi-classification approach leveraging multi-scale residual neural networks and multi-channel data fusion.

Main Methods:

• ECG signal features were extracted and transformed into 2D images.
• A multi-scale residual neural network was trained on labeled arrhythmia datasets.
• The classification model was applied for automatic arrhythmia detection during exercise.

Main Results:

• The proposed method achieved a classification accuracy of 99.60%.
• The model demonstrated high accuracy and strong generalization capabilities.
• Successful automatic identification of arrhythmias during exercise was achieved.

Conclusions:

• The developed multi-classification method offers a highly accurate solution for automated arrhythmia detection.
• This approach has significant implications for the future development of wearable arrhythmia monitoring devices.
• The findings support the integration of advanced AI techniques in cardiovascular diagnostics.