Research On Networked And Non-invasive Blood Pressure Intelligent Monitoring System

Blood pressure is one of the important indicators to evaluate human health.Its abnormal fluctuation will cause serious harm to human body.In recent years,with the increasing prominence of unreasonable dietary structure,mental stress,irregular work and rest,blood pressure-related diseases,such as hypertension,cardiovascular and cerebrovascular diseases,are becoming more and more common,and tend to be younger,so it is particularly necessary to strengthen the continuous monitoring of blood pressure.Most of the existing non-invasive and continuous blood pressure measurement methods are photoelectric blood pressure measurement,which can be achieved by studying photoelectric volume pulse wave(PPG),electrocardiogram(ECG)and intravascular elastic model.However,there are some differences in individual characteristics between different human bodies,so it is necessary to establish mathematical models between physiological signals and vascular elastic chambers for different populations,which increases the complexity of blood pressure measurement.In order to solve this problem,this paper proposes a blood pressure correction algorithm based on machine learning model,which combines the characteristics of PPG,ECG and individual features.Combining with the background and knowledge of network transmission,a network-based non-invasive blood pressure intelligent monitoring system based on human buttocks is designed.Firstly,in order to improve the quality and processing efficiency of physiological signals,an extraction algorithm based on high quality signals is proposed in this paper.Firstly,the pre-processed physiological signals are divided into several parts according to the characteristic points.Then,through statistical analysis,the signals are judged from two aspects: rough judgment of waveform morphological stability and accurate judgment of waveform morphological accuracy,so as to extract the high-quality signals from the signals,thus avoiding abandoning the whole segment of the signals due to an abnormal signal,which affects the efficiency of subsequent processing.Secondly,by analyzing the relationship between blood pressure production mechanism and various physiological characteristics of human body,this paper proposes basic information based on user's gender,age,height,weight,presence or absence of hereditary high and low blood pressure,body mass index,and heart rate measured in real time.And the blood pressure estimation value processed by the high-quality physiological signal is taken as an input parameter,and the training data set is constructed by taking the measurement result of the mercury sphygmomanometer at the same time as a real value.Then,support vector regression model and Stochastic Forest regression model were established to predict blood pressure,and genetic algorithm and grid search method were used to optimize the parameters of the model.Through a lot of experiential learning and experimental comparison,it is concluded that each evaluation index of blood pressure prediction results of Stochastic Forest regression model is superior to that of support vector regression model,so the Stochastic Forest regression model is finally selected as the blood pressure correction model in this monitoring system.Finally,the implementation of the whole network non-invasive blood pressure intelligent monitoring system is elaborated in detail from both hardware and software aspects.The hardware part is mainly the combination of network communication module and signal acquisition module.In software aspect,based on Visual Studio 2016 as software development platform,a set of multi-physiological parameters information processing system is developed.Its main functions include data communication,real-time display of physiological signals,display of physiological parameters processing results,query of historical information and so on.