Continuous blood pressure estimation using PPG and ECG signal

Abstract

Continuous blood pressure monitor can detect the potential risk of cardiovascular disease and provide a gold standard for clinical diagnosis. The features extracted from photoplethysmography (PPG) and electrocardiogram (ECG) signals can reflect the dynamics of cardiovascular system. In this paper, 39 features are extracted from PPG and ECG signals and 10 features are chosen by analyzing their correlations with blood pressure. Several machine learning algorithms are used to predict the continuous and cuff-less estimation of the diastolic blood pressure and systolic blood pressure. The results shows that compared with linear regression and support vector regression methods, the artificial neural network optimized by genetic algorithm gives a better accuracy for 1 h prediction under Advancement of Medical Instrumentation and the British Hypertension Society standard.