Prediction Of Human Blood Pressure Based On Wavelet Analysis

Blood pressure is an important physiological signs of the human body,reliable and effective blood pressure prediction is of great significance in medical decision-making.In recent years,cardiovascular diseases such as hypertension and cerebral hemorrhage have seriously threatened human health.In clinical,blood pressure is an important indicator of the diagnosis of hypertension,acute hypotension,and whether cardiovascular function is normal.So the effective of prediction arterial blood pressure can provide a decision-making information and early warning effect for health care workers,so that medical staff have valuable time to check the cause,and make the corresponding treatment plan,so as to avoid further deteroration of the disease.Meanwhile,the Internet,Internet of things and cloud computing and other technologies for the rapid development of human blood pressure prediction to provide technical support.In this paper,the human arterial blood pressure is taken as the research object,and the time series model is used to predict it.Firstly,based on the artificial neural network,analysis and selection of three kinds of artificial neural networks are widely used in the field of prediction: BP neural network prediction method,Elman neural network prediction method and BP neural network prediction method based on genetic algorithm optimization,these three methods are used to establish the blood pressure time series prediction model,and the prediction model of these three methods is analyzed and the improvement scheme is discussed.And then proposes a combination forecasting model based on wavelet analysis and BP neural network,the blood pressure time series is reconstructed by wavelet decomposition and single reconstruction respectively.The signal after decomposition is analyzed respectively,so that the BP neural network prediction model is established for each layer of the reconstructed wavelet signal,and the final prediction value of blood pressure is obtained by the prediction of each component.Finally,combined with the characteristics of wavelet decomposition and reconstructed signal,it is found that the high frequency signal of wavelet decomposition and reconstruction is a stationary signal,and the low frequency signal is still nonstationary signal,this paper proposes the based on wavelet analysis of BP neural network and ARMA combination forecasting model for this fueature,the prediction model of high frequency signal is reconstructed,the ARMA model is used to predict high frequency components,and the low frequency trend component is still predicted by BP neural network model.Experiments show that the improved combined prediction model can achieve better predictive effect on blood pressure prediction.In this paper,three artificial neural network models established by time series prediction method have achieved some effect in human blood pressure prediction,but it is not ideal.By introducing the wavelet multiresolution analysis,the Mallat algorithm is used to decompose the blood pressure time series into different frequency component components,so that the information of each layer is more single,and the prediction model can approximate the changes of human blood pressure and reach the reliable prediction requirement.The ARMA prediction model is established for the high frequency signal,and the BP neural network prediction model is still used in the low frequency signal.The combined prediction model further improves the accuracy of human blood pressure prediction.The combined forecasting model proposed in this paper provides a reliable and effective method for predicting human blood pressure.However,there are some shortcomings in this paper,the model does not consider the factors that affect the changes in blood pressure,in addition,the sample data may exist abnormal data on the prediction effect of the model,need to use a more complete data processing methods.