Blood Pressure Estimation Using PPG And Machine Learning

Abstract

Uninterrupted blood pressure tracking is indispensable for timely identification of cardiovascular conditions. Conventional pneumatic cuff instruments, while clinically validated, impose practical constraints that render them unsuitable for sustained use—primarily discomfort and the inability to yield readings between scheduled intervals. The present work introduces a cuffless, non-invasive framework built around Photoplethysmography (PPG) waveforms and supervised regression algorithms. Salient morphological and temporal features are derived from raw PPG recordings and subsequently employed to train and evaluate a suite of machine learning regressors for predicting both systolic and diastolic pressure. Experimental findings confirm competitive accuracy, and the integration of a browser-accessible monitoring interface positions the system as a practical candidate for wearable and telehealth deployments.