Research On Blood Pressure Prediction Method Based On Support Vector Regression Model

Hypertension is a potential health hazard,which is directly manifested by the blood pressure value of the human body.The traditional blood pressure measurement technology is mainly based on intermittent measurement,which will cause discomfort to people.In recent years,with the rapid development of science and technology,acquiring human physiological signals from wearable smart devices and then building a blood pressure prediction model has become a new blood pressure measurement method.However,this new blood pressure measurement method faces some problems and challenges,such as insufficient physiological signal data sources,noise pollution,and insufficient accuracy of blood pressure prediction methods,which limit its further development.In view of the above problems,deeply study on blood pressure prediction is carried out.The main research work and the achievements obtained are as follows:(1)Physiological signal data extraction.Acquire the permission to use the MIMIC III database of physiological signals of the human body,and extract the physiological signal data required for the study of blood pressure prediction methods from the database,mainly including Photoplethysmography(PPG)and Arterial Blood Pressure(ABP)signal data.(2)Study on noise pollution treatment methods of PPG signal.PPG signal is caused by noise pollution due to external light sources,movement,breathing,and other electromagnetic factors,which have a great impact on the accuracy of blood pressure prediction.Therefore,the noise should be eliminated before it can be used to predict blood pressure.To solve these problems,a new data filtering method BW for PPG signals is proposed.In this method,only Butterworth filter is used to deal with the noise pollution of PPG signal.First,butterworth filter is used to eliminate baseline drift of PPG signal,and then the high-frequency interference of PPG signal itself and caused by baseline drift are eliminated by this filter.Experimental results show that this method can effectively eliminate the noise pollution of PPG signal.(3)Study on methods of blood pressure prediction.To solve the problem of blood pressure prediction,a blood pressure prediction method BW-SVR based on Support Vector Regression(SVR)is proposed.First,the BW method proposed in the paper is used to filter the PPG signal and the signal is divided into periods.Then,the features of the filtered original PPG signal,as well as the related features of its first and second derivatives are extracted.Finally,the SVR algorithm is used to build a model for predicting blood pressure.The experimental results show that the BW-SVR method is superior to the existing methods in Mean Absolute Error(MAE)and Root Mean Square Error(RMSE),and at the same time,there is a high consistency between the predicted and observed values.The prediction error results meet the AAMI standard for blood pressure measurement.