Beyond Cuffs and Needles: Exploring Photoplethysmography-Based Machine Learning for Blood Pressure Estimation
Beyond Cuffs and Needles: Exploring Photoplethysmography-Based Machine Learning for Blood Pressure Estimation
Miften Sabar,Firas Haifaa,Dwaish
TLDR
The RF model outperformed the DT and SVM models with nearly 98%, 98% accuracy and F-score respectively, demonstrating its ability to capture complex non-linear relationships between PPG features and BP.
Abstract
This study explores the use of machine learning techniques to classify blood pressure (BP) based on Photoplethysmography (PPG) signals. Three popular algorithms - decision trees (DT), random forests (RF), and support vector machines (SVM) - were evaluated. The RF model outperformed the DT and SVM models with nearly 98%, 98% accuracy and F-score respectively , demonstrating its ability to capture complex non-linear relationships between PPG features and BP. It also showed robust performance in the presence of noise and variations in input PPG signals, making it a promising choice for real-world BP monitoring applications.