Learning-Based Model for Central Blood Pressure Estimation using Feature Extracted from ECG and PPG signals

Area of Science:

• Biomedical Engineering
• Cardiovascular Physiology
• Medical Signal Processing

Background:

• Hypertension pre-detection typically relies on Brachial Artery Blood Pressure (BABP), which is an indirect measure of Central Blood Pressure (CBP).
• Current CBP measurement is invasive, requiring cardiac catheterization, and cuffless estimation methods for CBP are underdeveloped.
• Continuous, beat-by-beat, cuffless CBP estimation has not been previously proposed.

Purpose of the Study:

• To develop and validate a novel cuffless, beat-by-beat technique for estimating Central Blood Pressure (CBP).
• To explore the use of Electrocardiogram (ECG) and Photoplethysmograph (PPG) signals for non-invasive CBP monitoring.
• To improve upon existing methods by incorporating advanced signal processing and feature extraction.

Main Methods:

• A cuffless, beat-by-beat CBP estimation model was developed using linear regression analysis.
• Features were extracted from continuous ECG and PPG signals, including 30 supplementary features beyond conventional pulse transit time (e.g., ST-interval, Psystolic peak interval) using Haar wavelet and modulus maxima.
• Feature selection utilized a wrapper technique, followed by dimensionality reduction with Principal Component Analysis (PCA). Beat segregation was achieved through iterative constraints and backtracing.

Main Results:

• The developed model accurately estimated Systolic CBP with a validation error of 0.109±2.37 mmHg.
• Diastolic CBP was estimated with a validation error of 0.031±2.102 mmHg.
• The study validated the technique on 33 cardiac catheterization lab patients.

Conclusions:

• A novel, non-invasive, cuffless, beat-by-beat method for CBP estimation using ECG and PPG signals has been successfully demonstrated.
• The incorporation of advanced signal processing and supplementary features significantly enhances the accuracy of CBP estimation.
• This technique holds potential for intermittent and continuous monitoring of central blood pressure, offering a valuable alternative to invasive methods.