Data-Driven Automated Cardiac Health Management with Robust Edge Analytics and De-Risking.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Internet of Things (IoT)

Background:

• Remote and automated healthcare management offers significant potential for improving patient prognosis.
• The Internet of Things (IoT) provides an ecosystem for developing scalable healthcare solutions.
• Cardiac health monitoring is crucial, and traditional methods can be enhanced with data-driven approaches.

Purpose of the Study:

• To demonstrate the clinical efficacy of data-driven techniques for cardiac health management using machine learning.
• To utilize phonocardiogram (PCG) signals for robust cardiac health assessment.
• To implement a privacy-preserving, automated cardiac screening system at the edge.

Main Methods:

• Employed machine learning methods with selected signal processing features on phonocardiogram (PCG) data.
• Developed a shallow classifier with three features for efficient edge gateway processing.
• Integrated differential privacy techniques for de-risking sensitive healthcare data management.

Main Results:

• Achieved close to 85% accuracy on publicly available MIT-Physionet PCG datasets.
• Outperformed relevant state-of-the-art algorithms in cardiac health classification.
• Demonstrated the feasibility of on-demand, automated cardiac health screening with minimized privacy risks.

Conclusions:

• Data-centric approaches using machine learning on PCG signals offer superior clinical utility for cardiac health management.
• Edge computing enables efficient, real-time analysis of physiological signals.
• Differential privacy is essential for protecting sensitive healthcare data in IoT environments, ensuring secure automated screening.