A Study on Arrhythmia via ECG Signal Classification Using the Convolutional Neural Network.

Area of Science:

• Medical technology
• Artificial Intelligence
• Cardiology

Background:

• Cardiovascular diseases (CVDs) are the leading global cause of mortality.
• Electrocardiogram (ECG) analysis is crucial for diagnosing CVDs but requires significant expert resources.
• Current machine learning methods for ECG analysis often involve manual feature extraction and complex models.

Purpose of the Study:

• To propose a robust and efficient deep learning model for classifying heartbeat types from ECG data.
• To address the limitations of existing machine learning approaches in ECG analysis, such as manual feature engineering and long training times.
• To improve the accuracy and efficiency of automated ECG interpretation for clinical practice.

Main Methods:

• Development of a 12-layer deep one-dimensional convolutional neural network (CNN).
• Classification of five micro-classes of heartbeat types using the MIT-BIH Arrhythmia database.
• Application of a wavelet self-adaptive threshold denoising method for data preprocessing.

Main Results:

• The proposed 12-layer CNN model demonstrated superior performance compared to traditional methods like BP neural networks and random forests.
• The model achieved higher accuracy, sensitivity, robustness, and anti-noise capability in classifying heartbeat types.
• Experimental results indicate significant improvements in automated ECG analysis.

Conclusions:

• The developed deep learning model offers an effective and efficient solution for ECG-based heartbeat classification.
• This approach can significantly reduce the reliance on manual expert analysis, thereby conserving medical resources.
• The findings suggest a positive impact on clinical practice through enhanced diagnostic capabilities.