Blood Pressure Estimation Based on the Fusion of Features Extracted from Photoplethysmography and Electrocardiography Signals

In this paper a novel noninvasive procedure for blood pressure estimation according to the fusion of features extracted from photoplethysmography (PPG) and electrocardiography (ECG) signals is presented. The main goal of the suggested approach is to improve cardiovascular monitoring accuracy by exploiting the complementary information provided by these two physiological signals. In the proposed method PPG and ECG signals are first preprocessed, and a set of ۲۱ blood pressure related physiological features is extracted. These features are then used as predictor variables (inputs) for a set of multivariate linear regression models to estimate blood pressure values. To assess the performance of the presented method, a dataset consisting of multiple signal recordings collected from different patients is utilized. Data preprocessing is carried out through principal component analysis (PCA) while feature dimensionality reduction and selection of the most informative features are achieved through stepwise linear regression. Finally three features exhibiting the highest correlation with blood pressure are identified and selected as the core features of the suggested approach. The effectiveness of the suggested approach is compared with existing methods, and the obtained results demonstrate that the suggested model achieves acceptable accuracy and satisfactory efficiency in blood pressure estimation.

کلیدواژه ها:

blood pressure ، Photoplethysmography ، Electrocardiography ، Multivariate linear regression ، Principal component analysis ، Stepwise Linear Regression