Development and Validation of Embedded Device for Electrocardiogram Arrhythmia Empowered with Transfer Learning.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Cardiology

Background:

• Internet of Things (IoT) and cloud computing have enhanced healthcare data management and disease investigation.
• Electrocardiogram (ECG) is crucial for diagnosing heart conditions.
• Previous machine learning techniques for ECG analysis were feature-based and less accurate than transfer learning.

Purpose of the Study:

• To develop and validate an embedded device model (DVEEA-TL) for accurate ECG arrhythmia diagnosis using transfer learning.
• To improve the accuracy of heart abnormality detection compared to existing methods.

Main Methods:

• Development of the DVEEA-TL model, integrating hardware and software components.
• Creation of a fused dataset combining Kaggle and real-time healthy/unhealthy ECG data.
• Application of the AlexNet transfer learning approach for ECG signal analysis.

Main Results:

• The DVEEA-TL model achieved high accuracy in diagnosing heart abnormalities.
• Achieved 99.9% accuracy during the training stage and 99.8% during the validation stage.
• Demonstrated superior performance compared to previous research in ECG-based heart abnormality detection.

Conclusions:

• The DVEEA-TL model offers a reliable and highly accurate approach for ECG arrhythmia diagnosis.
• The integration of transfer learning and a fused dataset significantly enhances diagnostic accuracy.
• This model represents a reliable advancement in automated heart condition monitoring.