Machine Learning Methods for Real-Time Blood Pressure Measurement Based on Photoplethysmography

This paper presents real-time blood pressure (BP) measurement methods based on Photoplethysmography (PPG) signal. One feature vector encompassing eight features from PPG signal is first extracted. Based on feature vector, various machine learning methods are used to estimate BP. The accuracy of different methods is evaluated on Queensland Vital Signs Dataset. Random Forest achieves the best performance in terms of mean absolute difference (MAD) and standard deviation (STD) of error. MAD ± STD of 4.21 ± 7.59 mmHg for SBP estimation and 3.24 ± 5.39 mmHg for DBP estimation are achieved. Grade A is obtained according to the British Hypertension Society protocol (BHS). Meanwhile, the proposed method meets the Advancement of Medical Instrumentation (AAMI) standard

Keywords: Random Forest Regressor, Decision Tree Regressor, Bagging Regressor, XGBoost, Gradient Boosting Regressor CatBoost Regressor K Neighbors Regressor, SVR, Extra Tree Regressor and StackingRegressors.