Research On Blood Pressure Detection Algorithm Based On Pulse Wave Of Fingertip PPG Signal

In recent years,with the increasing living standards of people in China,the prevalence of hypertension has also increased year by year.Hypertension,as a chronic disease that can be controlled but requires lifelong treatment,is difficult to detect in the early stages of the disease.At the same time,the traditional hospital blood pressure detection device lacks comfort and convenience,and can not meet people’s blood pressure monitoring requirements for individuals and families.The blood pressure detection method based on the photoplethysmograph method has the advantages of non-invasive,simple and comfortable.The wearable sphygmomanometer developed according to this method is increasingly used in daily life.Based on the fingertip photoplethysmograph acquisition system,this paper proposes two blood pressure algorithms based on ECG pulse method and single-channel accelerated pulse wave method.According to the extracted pulse wave characteristic parameters,the neural network is used to complete the segmented estimation of blood pressure.Specific research work includes:1.Realized blood pressure algorithm based on Pulse Wave Transit Time(PWTT).Through research on the the key design methods of extracting PWTT,two detection schemes were selected:One is to choose electrocardiogram(ECG)and photoplethysmograph(PPG)to extract the R point in ECG signal and the trough in PPG signal to calculate PWTT;second is to select single pulse wave signal,make a quadratic difference to obtain an accelerated pulse wave,and then select two feature points in the accelerated pulse wave to calculate PWTT.First,the wavelet transform method is used to pre-process the ECG and PPG signals.Secondly,based on mathematical morphology,extreme value method,fast Fourier transform and other methods,the peak and valley positions of the signal and various time-domain frequency-domain characteristic parameters are extracted.Respectively analyze the correlation between PWTT and blood pressure extracted by the two schemes and build a linear model.Finally,conduct a correlation study on the extracted feature parameters,and select appropriate time and frequency domain feature parameters to improve the blood pressure calculation formula.2.Constructed a neural network algorithm model to estimate the blood pressure interval.Through physiological theory,selected multiple characteristic parameters in PPG signal for machine learning.The time domain characteristic parameters and frequency domain characteristic parameters of the pulse wave in the collected clinical data are respectively applied to the neural network model to realize the judgment of the blood pressure interval.This paper designs experiments to evaluate the two blood pressure detection algorithms.Measured by the Bland-Altman analysis method,the average difference between the two algorithms to detect the blood pressure value is not higher than±5mmHg,which meets the requirements of the AAMI average difference of less than 5 mmHg and standard deviation of less than 8 mmHg.The correctness and feasibility of the testing scheme in this paper have been verified.