Research On Noninvasive Continuous Blood Pressure Measurement Based On Elman Neural Network

The standard of blood pressure can directly reflect the health of human cardiovascular system.Blood pressure should be used to reflect the patient's vital signs for critically ill patients who are in the operation in medical treatment.It also plays a vital role in the prevention of cardiovascular disease in the subjects,In family health care.However,the main clinical methods of blood pressure measurement are mainly intermittent.As the fluctuation of blood pressure varies every moment,the method of continuous blood pressure measurement is of great significance in both clinical and family health care.At present,two methods of continuous blood pressure measurement can be divided into non-invasive and invasive methods.The measurement accuracy of invasive continuous blood pressure measurement is high.Because of its complicated operation may infect the patients so the function can't be universally recognized by the industry.In the method of noninvasive continuous blood pressure measurement,it can be divided into pulse wave velocity method(PWV)and pulse wave characteristic parameter method.Pulse wave propagation velocity method: from ECG signal and synchronous pulse signal we can get the propagation time between the peak point of the electrocardiogram and the peak value point of the pulse wave.And then the velocity of pulse propagation(PWV)can be calculated.A linear blood pressure regression equation is established by the obtained PWV,and the continuous estimation of blood pressure is achieved.Because of the nonlinear relationship between blood pressure and pulse velocity obtained from blood volume tracing(PPG)signal,the calculation error is relatively large during the real-time measurement of blood pressure.Pulse wave characteristic parameter method: which is to extract feature points from the corresponding period of each pulse wave,such as the relative height of the pulse wave and dicrotic notch(h/H),the relative height of the dicrotic wave(g/H),systolic time,relaxation time.By regression analysis of the above features,a regression equation is established to achieve continuous measurement of blood pressure.This method only applies the corresponding relationship between the characteristic point of a particular time and the blood pressure at this time.But it ignore the impact of the pulse characteristics on the current blood pressure at the last moment,so the accuracy of its measurement will be affected.The current non-invasive continuous blood pressure measurement method can not accurately obtain the continuous blood pressure measurement value.In order to overcome the limitations of the continuous blood pressure measurement,a continuous blood pressure measurement method based on pulse feature points and Elman neural network is proposed in this paper.The method take the single channel pulse signal as the research object,using the time-related features of Elman neural network,a mathematical model is established,and the measurement of continuous blood pressure is completed.The synchronous pulse signal and the continuous blood pressure data in the MIMIC physiological parameter database are used as the research object in this paper.First we eliminate the noise of pulse signal,extract feature point and other correlation preprocessing.Using the pulse characteristic parameters of the previous moment and the current time,we construct the matrix as the training set feature input of Elman neural network.We select the SBP and DBP at current time as the output of the training model,finally we get a model of continuous blood pressure measurement.Secondly,in order to verify the accuracy of Elman neural network continuous blood pressure measurement method,we select LR linear regression continuous blood pressure measurement method and BP neural network continuous blood pressure measurement method to compare with the method used in this paper.Finally,MATLAB is used to build GUI human-machine interaction system,and Elman neural network is used as the core algorithm,which initially realizes continuous measurement of blood pressure.