Continuous Blood Pressure Estimation Based On Time And Frequency Domain Of Pulse Wave

Blood pressure is one of the important indicators of human health.Abnormal blood pressure can cause burdens and even damage to various organs of the body.In addition,blood pressure is an important reference index for whether or not a lesion occurs in the cardiovascular,cardiac,and the like.Traditional blood pressure measurement will apply a cuff to the superficial artery such as the upper arm,afterwards,inflate the cuff and then slowly deflate.Finally,the blood pressure is obtained according to the pressure returned by the pressure sensor,but the blood pressure is caused by the pressure given by the cuff,which makes people feel uncomfortable and paralyzed.The blood pressure of the same individual changes during the day is about 5mmHg for the changes of the individual's mood and different pathological environment result in that the intermittent blood pressure estimation do can't meet the individual's different application conditions.In order to avoid the drawbacks caused by traditional methods,this paper designs a continuous blood pressure estimation system based on neural network dual-domain fusion model.The pulse wave waveform carries many physiology information of the human body,and it has a great correlation with the elasticity of blood vessels,the viscosity of blood,etc.,which directly affect blood pressure.In this regard,that is why this paper uses pulse waves as a signal source of blood pressure estimation.The specific study are as follows.(1)Design the appropriate hardware circuit to accomplish the following two tasks ?1 Place the piezoresistive sensor in the radial artery and acquire a clear pulse wave with a certain pressure ?2 The pulse wave will be processed by hardware filtered and then transmitted to the host computer after A/D sampling.(2)To perform a software filter for the pulse wave containing noise,and the period segmentation after obtaining a clean pulse wave.(3)Extract respectively six feature points in the time domain and the frequency domain as the input parameters of the neural network,and then use the commercial sphygmomanometer as a label to obtain systolic blood pressure(SBP)and diastolic blood pressure(DBP)through supervised learning.The above is the methodology and the following research results were obtained by the research of the continuous blood pressure estimation system of this paper.A: Piezoresistive sensor can emerge more details of pulse wave than photoplethysmography(PPG).It is found that the piezoresistive sensor can directly acquire the pulse wave signal relative to PPG,and the opposite is that PPG indirectly acquires the pulse wave signal through the absorption rate of light by the blood.B: Piezoresistive sensors need appropriate pressure to obtain excellent pulse wave signals.When acquiring pulse wave signals,different pressures of different sensors will detect different forms of pulse waves,which has an extremely large errors for the estimation of continuous blood pressure.Therefore,a suitable pressure is particularly significant.In this paper,for the pulse wave measurement is performed on different individuals by gradient pressure,the optimum pressure curve is obtained by using the body mass index(BMI)to perform the third-order polynomial fitting.C: The dual-domain fusion features can describe the blood pressure and cardiovascular function of the body more than the single domain's features.Each peak points of the time domain corresponds to the related physiological activity,but extracting accurately the features under external disturbance environment is difficult.Fortunately,the frequency domain features can compensate for the shocks caused by the time domain.The results show that the continuous blood pressure estimation based on time and frequency domain of pulse wave can achieve the continuous blood pressure estimation with higher precision.The error range of SBP and DBP is 0.18 T 4.22 mmHg,0.027 T 3.67 mmHg respectively.