A Novel Non-Invasive Cuff-Less Method for Continuous Blood Pressure Estimation Using Machine Learning and Photoplethysmography

Hypertension requires timely monitoring to prevent complications like strokes and heart failure. This study introduces a non-invasive, cuff-less method for continuous blood pressure (BP) estimation using machine learning (ML) and photoplethysmography (PPG) signals. Our approach reveals BP fluctuations by examining the correlation between arterial BP (ABP) and PPG signals. Validation with standard metrics shows that a 16-second segmentation technique and Support Vector Machine (SVM) algorithm achieve mean absolute errors (MAE) of 10.54 for systolic BP (SBP) and 5.28 for diastolic BP (DBP), with R-squared values of 0.59 and 0.34, respectively. This method offers a practical and superior alternative for BP monitoring in healthcare applications.