An effective feature extraction method based on GDS for atrial fibrillation detection.

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Atrial fibrillation (AF) is a prevalent and dangerous heart arrhythmia.
• Accurate and timely AF detection is crucial for patient outcomes.
• Current AF detection methods often rely on complex signal processing, posing challenges for variable ECG data.

Purpose of the Study:

• To develop a simplified feature extraction method for automatic atrial fibrillation detection.
• To address the limitations of existing complex AF detection techniques.
• To enhance the accuracy and efficiency of AF diagnosis using electrocardiogram (ECG) signals.

Main Methods:

• A novel feature extraction technique using gradient set (GDS) was developed.
• GDS features were derived from ECG segments.
• Statistical distribution and information quantity features of GDS were calculated for classifier input.

Main Results:

• The proposed GDS method demonstrated simple calculations and noise tolerance.
• The method showed high adaptability across various classifiers.
• Optimal performance was achieved with a designed deep neural network (DNN) classifier.

Conclusions:

• The gradient set (GDS) feature extraction method is effective for atrial fibrillation detection.
• This approach offers a robust and adaptable alternative to complex existing methods.
• The GDS method is a suitable choice for feature extraction in AF detection systems.