Self-supervised Learning with Demographic Information for Cuffless Blood Pressure Estimation.

Area of Science:

• Biomedical Engineering
• Physiological Monitoring
• Machine Learning in Healthcare

Background:

• Photoplethysmography (PPG) offers convenient, continuous physiological monitoring via wearables.
• Cuffless blood pressure (BP) estimation using PPG is a key application, but accuracy is challenged by signal complexity and individual variability.
• Existing self-supervised learning methods for PPG-based BP estimation often overlook crucial demographic factors like age, gender, height, and weight.

Purpose of the Study:

• To develop and evaluate a novel self-supervised learning framework for cuffless blood pressure estimation.
• To integrate demographic information (age, gender, height, weight) into the PPG-based BP estimation model.
• To enhance the accuracy of blood pressure estimation by accounting for individual physiological differences.

Main Methods:

• A new self-supervised learning algorithm was proposed for blood pressure estimation.
• The method incorporated demographic data alongside PPG signals.
• Model performance was validated using the public PulseDB dataset.

Main Results:

• The proposed method achieved Mean Absolute Errors (MAE) of 5.41 mmHg for systolic blood pressure (SBP) and 2.27 mmHg for diastolic blood pressure (DBP) when using PPG alone.
• These accuracy metrics demonstrate superior performance compared to recent state-of-the-art methods.
• The integration of demographic information is shown to be beneficial for improving BP estimation accuracy.

Conclusions:

• The developed self-supervised learning method effectively estimates blood pressure using PPG and demographic data.
• Incorporating demographic information significantly enhances the accuracy of cuffless blood pressure estimation.
• This approach represents a promising advancement for non-invasive, wearable blood pressure monitoring.