Convolutional Neural Networks for patient-specific ECG classification.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Medicine
• Cardiology

Background:

• Electrocardiogram (ECG) analysis is crucial for diagnosing cardiac arrhythmias.
• Current methods may lack patient-specificity and real-time processing capabilities.
• Developing efficient and accurate ECG classification systems is an ongoing challenge.

Purpose of the Study:

• To develop a patient-specific ECG classification and monitoring system.
• To integrate feature extraction and classification using a unified deep learning model.
• To enable fast and accurate detection of cardiac abnormalities like ectopic beats.

Main Methods:

• An adaptive 1D Convolutional Neural Network (CNN) architecture was employed.
• A dedicated CNN was trained for each patient using limited common and patient-specific data.
• The system fuses feature extraction and classification into a single learning process.

Main Results:

• The proposed patient-specific CNN achieved superior classification performance.
• The system demonstrated high accuracy in detecting ventricular ectopic beats (VEB) and supraventricular ectopic beats (SVEB).
• The approach allows for efficient classification of long ECG records (e.g., Holter).

Conclusions:

• Patient-specific adaptive CNNs offer a fast and accurate solution for ECG analysis.
• The system is suitable for real-time ECG monitoring and early alert systems on wearable devices.
• This approach enhances the detection of critical cardiac events like VEBs and SVEBs.