Improved Heart Rate Tracking Using Multiple Wrist-type Photoplethysmography during Physical Activities

Area of Science:

• Biomedical Engineering
• Signal Processing
• Wearable Technology

Background:

• Photoplethysmography (PPG) signals from wearables are prone to motion artifact (MA) during exercise, challenging accurate heart rate (HR) estimation.
• Existing methods struggle with MA, limiting the reliability of HR monitoring in real-world physical activities.

Purpose of the Study:

• To develop and validate a novel framework for accurate HR estimation from PPG signals in the presence of MA.
• To leverage accelerometer (ACC) data in conjunction with dual-lead PPG signals to mitigate MA effects.

Main Methods:

• A moving time window approach segments PPG and ACC signals.
• Joint sparse spectrum reconstruction attenuates MA by subtracting ACC spectrum frequencies from PPG spectrum frequencies.
• HR is estimated from the dominant frequency in the cleansed PPG spectrum and refined using spectral band powers.

Main Results:

• The proposed method achieved an average absolute HR error of 1.15 BPM (SD: 2.00 BPM) and an average absolute percentage error of 0.95% (SD: 1.86%).
• Validation on the 2015 IEEE Signal Processing Cup dataset demonstrated superior accuracy compared to previous methods.
• The framework effectively reduces motion artifact, enhancing the reliability of HR estimation.

Conclusions:

• The proposed framework offers a robust solution for accurate HR estimation from wearable PPG during physical exercise.
• Combining dual-lead PPG with ACC data and sparse spectrum reconstruction is effective in overcoming motion artifact challenges.
• This approach significantly improves the accuracy and reliability of HR monitoring in wearable sensing applications.