Peak-Independent Cuffless Blood Pressure Monitoring Using a Smart Sock: The Role of Temporal Lag Modeling in...

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Cardiovascular Monitoring

Background:

• Continuous blood pressure (BP) monitoring is crucial for healthcare but challenging with current cuff-based devices.
• Wearable systems require non-invasive, long-term BP monitoring solutions.

Purpose of the Study:

• To develop a cuffless BP estimation method using a Smart Sock platform and photoplethysmography (PPG).
• To explore peak-independent features and temporal modeling for enhanced BP monitoring.

Main Methods:

• Utilized single-site PPG signals from 60 participants, processed in 30s windows.
• Extracted statistical and information-theoretic features, including derivatives, and applied temporal lag modeling (5-15s).
• Employed multiple regression models with leakage-safe cross-validation for BP estimation.

Main Results:

• Subject-independent estimation achieved errors of 8.60 mmHg (systolic) and 6.42 mmHg (diastolic).
• Post-calibration monitoring yielded significant improvements with mean absolute errors of 1.3-1.7 mmHg and R² > 0.90.
• Demonstrated the efficacy of peak-independent features and temporal context modeling.

Conclusions:

• Foot-based PPG, combined with advanced feature engineering and temporal modeling, enables accurate continuous personalized BP monitoring post-calibration.
• Subject-independent BP estimation remains a significant challenge.
• The Smart Sock platform offers a comfortable and effective alternative for long-term BP management.