The Research Of High-Resolution Continuous Blood Pressure And Blood Pressure Variability Measurement Based On Multi-Parameters Fusion

Blood pressure and blood pressure variety are both independent influencing factors for cardiovascular events,real-time monitoring of the two physiological indicators is of great significance to human health,so this paper is to probes into how to obtain accurate non-invasive continuous blood pressure simply,and it can provide a viable solution for wearable equipment.A methodology for continuous cuffless blood pressure measurement based on multi-parameters fusion and a feasible modeling method for blood pressure variety are proposed,and they have many advantages,such as easy integration,high precision and so on.Using data mining for feature selection before modeling can find new factor strongly correlated to blood pressure and reducing complexity of the blood pressure model.Individual modeling based on the same feature combination can simplify the process and improve estimation accuracy for blood pressure.The first part of the research is continuous blood pressure modeling,a new method with high efficiency is proposed to detect feature points in ECG and PPG,and then the features are extracted.The correlation analysis and genetic algorithm are used to select the features strongly associated with BP and four features are obtain.Individualized BP estimation models are built by using multiple linear regression and support vector regression and tested.The result shows that the model built with SVR based on the four features has higher precision,and the mean difference and standard deviation between the estimated value and reference value are(0.0004±3.83)mmHg for SBP and(-0.001±2.652)mmHg for DBP,which are agreement with AAMI standard,so we can conclude that the method can reducing complexity of the model by a big margin while high precision is ensured.The second part is the research of blood pressure variety modeling and the exploration of the influence on blood pressure variety of short-time excesice.Individualized estimation models of BPV is established based on the variety of PTT.The result shows that the mean Pearson's correlation coefficient between PTTV and BPV is 0.415 and the maximum one is 0.836,it demonstrates the feasibility of cuff-less continuous BPV monitoring.Then the indexes in time-frequency domain of BPV are calculated and compared before and after exercise,it turns out that the standard deviation,the root mean square successive difference and the low-frequency power of SBP is become much bigger after exercise.