Key Technology Study Of Wearable Core Body Temperature And Continuous Blood Pressure Measurements

Global healthcare systems are struggling with aging population,prevalence of chronic diseases,and the accompanying rising costs.In this background,wearable health monitoring for families and individuals has been studied widely by lots of researchers in recent years.The main research objects of the wearable physiological parameters include electrocardiography(ECG),blood oxygen saturation(SpO2),pulse rate(PR),blood pressure(BP)and body temperature(BT),etc.These physiological parameters,which are closely related to human health condition and disease protection,are worthy of continuous monitoring.The wearable physiological parameter monitoring systems(WPPMS)can fulfill the non-invasive measurement of these vital physiological parameters,and are suitable for daily health mangement.At present,some WPPMS have achieved the measurements of single/multi-lead ECG,SpO2,PR,cuff-based BP,surface BT,etc,and a few of them have also fulfilled the continuous BP measurement.It is worthy to note that cuff-based BP measurement can only monitor BP intermittently,and can not track continuous BP fluctuations.Besides,traditional continuous BP measurements have some drawbacks such as low precision and frequent calibration.Surface BT,which is easily affected by ambient temperature and sweat evaporation,can not reflect the deep temperature inside human body.Therefore,reliable methods of core body temperature(CBT)and continuous BP measurements are of great significance and are worthy to study.In order to solve the aforementioned problems,this dissertation developed a WPPMS and improved the methods of CBT and continuous BP measurements.Besides,a cardiovascular disease(CVD)classifier was also studied.The main contents of this dissertation are as follows:(1).Design and implementation of WPPMS:A WPPMS prototype was designed and implemented.This system was composed of BT,ECG,SpO2/PR,BP subsystems,which achieved the functions of CBT,multi-lead ECGs,SpO2,PR,cuff-based and continuous BP measurements.(2).Study of non-invasive CBT measurement:An improved CBT measurement based on the dual heat flux principle was proposed.Finite element simulation and analyses provide the theoretical guidance for the design of CBT probe.Besides,a reformative heat transfer medium composed of calcium carbonate powder and PDMS was developed to increase the thermal response of the probe.A adaptive filter was designed to suppress the influence of ambient temperature fluctuations.Experiments were carried out on hot plate and human body.Good measurement accuracy,quick thermal response speed,and high consistency were obtained.(3).Design and implementation of a SOI-based Ni thin-film temperature sensor:A SOI-based Ni thin-film temperature sensor was designed and fabricated.The image of the thermal-resistive strips and lead pads were developed by exposure to ultraviolet light,using a negative photoresist and vacuum evaporation.The design and manufacture process was preliminarily studied,which helped to pave the way for the integration of CBT probe in the future.(4).Study of continuous BP estimation:An improved continuous BP measurement based on the pulse transit time(PTT)was proposed.Compared with the conventional methods using Moens-Kortweg and Bramwell-Hill equations,this dissertation derived a proportional relationship between PTT-2 and BP from the Navier-Stokes equation and construct the BP models with hemodynamic covariates in addition to PTT.Univariate and multivariate regression analyses were conduct to select the model parameters.Regularization linear regression and recursive least square algorithm were adopted for model initiation and calibration.Compared with other two representative works in the MIMIC database study,the proposed method had higher measurement accuracy and less calibration frequency.Besides,this method was verified on human volunteers,good accuracy and consistency were also acquired.(5).Study of a CVD classifier:A classification method based on principal component analysis(PCA)and support vector machine(SVM)for CVD diagnosis was presented.After the signal processing and feature extraction of ECG,PPG and BP signals in the MIMIC database,PCA was adopted to find the main components of these features.A SVM with slack variables and a radial basis kernel function was utilized to build the classifier for preliminary diagnosis of CVDs.In the learning and training process of the classifier,a firefly algorithm was used to optimize the key parameters of the SVM.The 10-fold cross validation on the MIMIC database was conducted,and a classification accuracy of 95.8%for subjects with/without CVDs was obtained.This classification method can provide some guidance for the preliminary diagnosis whether people have CVDs or not.