ECG Classification Using an Optimal Temporal Convolutional Network for Remote Health Monitoring.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Cardiology

Background:

• Healthcare access remains unequal, particularly for rural populations, despite technological advancements.
• Remote healthcare solutions are crucial for bridging this gap and ensuring equitable access to medical services.
• Electrocardiogram (ECG) analysis is a fundamental diagnostic tool for identifying cardiac anomalies.

Purpose of the Study:

• To propose a novel machine learning (ML) architecture for ECG classification.
• To accurately diagnose five common heart diseases using ECG data.
• To develop a system suitable for remote health monitoring applications.

Main Methods:

• Implementation of a novel machine learning architecture based on temporal convolution networks (TCN).
• Utilizing dilated causal one-dimensional convolution on input heartbeat signals for feature extraction.
• Training and evaluation of the TCN model for multi-class ECG classification.

Main Results:

• The proposed TCN architecture achieved a high accuracy of 96.12% in classifying five heart diseases.
• An F1 score of 84.13% was obtained, indicating robust performance.
• The model demonstrated efficiency with a reduced parameter count (10.2 K), suggesting suitability for embedded systems.

Conclusions:

• The developed TCN-based ML model shows superior performance in ECG classification compared to existing methods.
• Its efficiency and accuracy make it a strong candidate for low-power, cost-effective remote health monitoring devices.
• This technology has the potential to significantly improve access to cardiac diagnostics in underserved areas.