Hypertension Detection with Machine Learning Classifiers Using PPG Signals
Authors :- Telangore, H., Sharma, M.
Publication :- Paradigm Shifts in Communication, Embedded Systems, Machine Learning, and Signal Processing. PCEMS 2024. Communications in Computer and Information Science, vol 2490. Springer, 2025.
Hypertension, a widespread and often asymptomatic cardio-vascular condition, poses a significant global health threat. Uncontrolled hypertension can lead to severe consequences, including heart attacks, strokes, kidney diseases, and heart failure. Early-stage detection of hypertension is paramount for several reasons. Firstly, the insidious nature of hypertension means individuals may be unaware of their condition until irreversible damage occurs. Secondly, early intervention allows for lifestyle modifications, medication, or other preventive measures to mitigate the progression of hypertension and reduce associated health risks. Precise detection is imperative for effective intervention. This study is dedicated to utilizing machine learning classifiers to identify hypertension through the analysis of Photoplethysmography (PPG) signals. Leveraging the Wavelet scattering for feature extraction enables the capture of both temporal and frequency domain information. The dataset, obtained from Guilin People Hospital and Guilin University of Electronic Technology in China, encompasses a diverse array of physiological recordings that mirror various hypertensive conditions. Employing Wavelet Scattering, acknowledged for its discriminative feature extraction capabilities, enhances the analysis of PPG signals. Various machine learning classifiers are employed to recognize hypertensive patterns based on the extracted features. Notably, k-Nearest Neighbors (KNN) exhibits exceptional performance, achieving a classification report: Accuracy 99.3%, Precision 99.2%, Recall 99.2%, F1-score 99.2%. This research aspires to contribute to the expanding realm of non-invasive methodologies for hypertension detection. Through the utilization of advanced signal processing and machine learning techniques, we aim to elevate the precision and efficiency of hypertension diagnosis, addressing a pivotal aspect of global cardiovascular health, all while ensuring the integrity of original content.