Design and implementation of a non-invasive and cuff-less arterial blood pressure monitoring system

Abstract

Hypertension or high blood pressure (BP) is one of the most common worldwide disease leading to heart attack or stroke. Continuous assessment of blood pressure level is key to diagnosing hypertension. In this study, we designed and tested a dedicated cuff-less monitoring system which estimates BP level without need for calibration. We obtained continuous measurements from 40 healthy subjects (30 males and 10 females) ranging from 20-30 years old. Our measurement protocol consisted of 15 minutes simultaneous electrocardiography (ECG) and photoplethysmography (PPG) within three sessions, i.e. rest, bicycle exercise, and recovery. From ECG and PPG signals, we obtained 34 candidate features from which up to 9 features were selected to estimate systolic and diastolic BP levels. We validate our results with three regression models such as linear regression, support vector machines (SVM) regression, and multilayer perceptron (MLP) to obtain the best results. The study provides a promising approach for modern cuff-less BP monitoring devices.