Non-Invasive Blood Pressure Estimation Enhanced by Capillary Refill Time Modulation of PPG Signals.

Area of Science:

• Biomedical Engineering
• Physiological Monitoring
• Machine Learning in Healthcare

Background:

• Non-invasive continuous blood pressure (CBP) estimation is crucial for managing cardiovascular health.
• Photoplethysmography (PPG) signals offer a promising, non-invasive method for CBP monitoring.
• Current PPG-based CBP estimation methods face accuracy challenges, necessitating novel approaches.

Purpose of the Study:

• To investigate the impact of capillary refill time (CRT) modulation on PPG signal quality.
• To enhance the accuracy of non-invasive continuous blood pressure (CBP) estimation using modulated PPG signals.
• To evaluate the performance of different machine learning models (ResNetCNN, LSTM, Transformer) with CRT-modulated PPG data.

Main Methods:

• Collected PPG data from 21 healthy participants during standardized CRT induction (9 N pressure for 15 s).
• Segmented PPG signals into 30-s intervals, comparing CRT-modulated and standard data.
• Validated three machine learning models (ResNetCNN, LSTM, Transformer) using leave-one-subject-out (LOSO) and non-LOSO cross-validation.

Main Results:

• CRT modulation significantly improved accuracy across all tested machine learning models.
• ResNetCNN demonstrated substantial gains, reducing MAE by up to 35.6% and MAPE by up to 40.6%.
• All models met AAMI criteria for medical device accuracy (mean error < 5 mmHg, SD < 8 mmHg).

Conclusions:

• CRT modulation is a highly effective strategy for enhancing PPG signal quality for CBP estimation.
• Machine learning models, particularly ResNetCNN, show significant improvements with CRT-modulated data.
• This approach holds strong potential for improving wearable CBP monitoring devices, especially in resource-limited settings.