Design of a Realtime Photoplethysmogram Signal Quality Checker for Wearables and Edge Computing.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Wearable Technology

Background:

• Photoplethysmogram (PPG) signals are crucial for monitoring physiological health parameters like heart rate and blood pressure.
• Accurate health parameter estimation relies heavily on high-quality PPG signals with minimal artifacts.
• PPG signals from consumer-grade devices (smartphones, wearables) often suffer from motion artifacts, necessitating quality assessment.

Purpose of the Study:

• To develop a computationally inexpensive PPG signal quality checker suitable for edge devices, such as smartwatches.
• To address the limitations of existing, computationally expensive machine learning and deep learning methods for PPG signal quality assessment in resource-constrained environments.

Main Methods:

• Design and implementation of a novel, edge-compatible algorithm for real-time PPG signal quality checking.
• Validation of the algorithm on diverse PPG datasets, including data from wearable, ICU, and medical-grade devices.
• Comparative analysis of algorithm performance, particularly in high-noise wearable scenarios.

Main Results:

• The proposed quality checker achieved a high F-score exceeding 0.92 in noisy wearable environments.
• Demonstrated significant improvement in the accuracy of heart rate calculations derived from smartphone PPG applications after applying the quality checker.
• The algorithm proved effective across various data sources, indicating robustness.

Conclusions:

• The developed edge-compatible PPG quality checker is effective for real-time signal assessment, particularly in challenging wearable settings.
• Implementing this quality checker enhances the reliability and accuracy of health parameter estimations from consumer-grade PPG devices.
• This approach offers a practical solution for improving PPG-based health monitoring on wearable and edge computing platforms.