Cuff-less blood pressure estimation using wrist photoplethysmography.

Area of Science:

• Biomedical Engineering
• Medical Devices
• Health Informatics

Background:

• The growing elderly population necessitates advanced healthcare solutions, particularly for managing chronic diseases.
• Current wearable medical devices often lack standardization and accuracy, hindering their clinical adoption.
• Reducing hospitalization through remote monitoring is crucial for managing healthcare costs and patient well-being.

Purpose of the Study:

• To develop and validate a system for accurate systolic and diastolic blood pressure (BP) measurement using only wrist photoplethysmography (PPG).
• To improve the accuracy of BP estimation from PPG signals through advanced signal processing and machine learning techniques.
• To address the limitations of current non-standardized and inaccurate wearable BP monitors.

Main Methods:

• Utilized photoplethysmography (PPG) signals acquired from the wrist.
• Applied machine learning algorithms to map morphological features of PPG signals to blood pressure values.
• Developed signal quality selection methods to enhance measurement accuracy.

Main Results:

• Achieved an accuracy improvement of up to 33.5% compared to methods without signal selection.
• Demonstrated a mean absolute error of 1.1mmHg in a personalized BP measurement scenario.
• Reported a mean absolute error of 8.7mmHg in an uncalibrated leave-one-out scenario.

Conclusions:

• Wrist-based PPG, combined with machine learning and signal selection, offers a viable method for accurate blood pressure monitoring.
• The developed system shows potential for improving the reliability of wearable medical devices for continuous BP tracking.
• This technology could significantly aid in managing chronic diseases and reducing healthcare costs through remote patient monitoring.