DDCNN: A Deep Learning Model for AF Detection From a Single-Lead Short ECG Signal

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Cardiology

Background:

• Wireless Body Sensor Networks enable convenient, real-time collection of single-lead electrocardiogram (ECG) data.
• Early detection of atrial fibrillation (AF) is a critical research area due to its prevalence and associated risks.

Purpose of the Study:

• To propose an effective deep learning method for detecting atrial fibrillation (AF) from short, single-lead ECG recordings.
• To enhance the performance and robustness of AF detection through data augmentation techniques.

Main Methods:

• A two-channel convolutional neural network (CNN) architecture was developed.
• Raw ECG signals were denoised, and heart rate (HR) values were extracted.
• A data augmentation method was employed to expand the training dataset and improve signal diversity.
• CNN processed ECG signals and HR values for feature extraction and classification.

Main Results:

• The proposed method demonstrated effectiveness in detecting AF from single-lead ECG data.
• Validation experiments confirmed the method's advantages compared to existing state-of-the-art approaches.
• Data augmentation significantly increased the diversity of training signals, improving model generalization.

Conclusions:

• The developed two-channel CNN combined with data augmentation offers a promising solution for automated AF detection.
• This approach facilitates early diagnosis and management of atrial fibrillation using readily available ECG data.
• The study highlights the potential of AI in analyzing wearable sensor data for cardiovascular health monitoring.