Study On Cuff-less And Continuous Blood Pressure Estimation And Application

Current researches showed that besides the level of blood pressure,the dynamic change of blood pressure is also a very important risk factor of cardiovascular and cerebrovascular diseases.The continuous blood pressure estimation method based on pulse transit time could assess blood pressure without using a cuff,which could realize the measurement of blood pressure anytime and anywhere without interfering with people’s daily life.It can not only provide the absolute level of blood pressure,but also provide the dynamic analysis of blood pressure,which has a broad application prospect in the family medical care service At present,this method estimates continuous blood pressure indirectly by calculating pulse transit time from ECG and PPG.However,this method is affected by factors such as vascular elasticity,and the estimation accuracy is limit The morphology of PPG reflects vascular elasticity information.In recent years,more and more studies have considered taking more indicators extracted from PPG and other indicators that can reflect vascular elasticity as covariables and incorporating them into the continuous blood pressure estimation model to reduce the impact of vascular elasticity on the accuracy of blood pressure estimation.In spite of this,the PPG features currently used is not enough to explain the corresponding relationship between PPG morphology and blood pressure,making the estimation accuracy of cuff-less and continuous blood pressure limit and difficult to meet the application requirementsTherefore,in order to reduce the factors such as vascular elasticity that affected the estimation accuracy of cuff-less and continuous blood pressure,and to improve the accuracy of blood pressure estimation,this study proposed a series of multi-order multi-variable PPG indicators that were extracted from the original,the first derivative,and the second derivative wave of PPG based on the effect of fluid dynamics as well as the vascular elasticity on PPG morphology.Specifically,the choice of the new PPG indicators was based on an assumption that hemodynamics of blood pulse would affect the rate of the PPG wave to ascend and descend.In details,PPG wave is a mix of both the forward wave and the reflected wave,which is influenced by the interaction of cardiac output and total peripheral resistance.Higher blood pressure induced by higher cardiac out and/or total peripheral resistance would induce greater forward wave and reflect wave,thus would affect the rate of the PPG wave to ascend and descend.It would influence the rising amplitude and slope of the original PPG wave,time interval,and so the area under the ascending/descending branches of the original PPG wave.Therefore,these indicators of PPG wave are not only related to peripheral arteries,but also related to cardiac output,which are two factors determining blood pressure value.The inclusion of these PPG indicators in the blood pressure estimation model is expected to improve the accuracy of blood pressure estimation.In addition,blood vessels stiffen with age,which could increase the amplitude of reflected wave thus enhance systolic pressure.This makes the PPG waves appear sharper and narrower.Therefore,the analysis of features such as time interval between ascending branch and descending branch of PPG wave can also reflect the impact of vascular elasticity on blood pressure.In addition,we also found some PPG indicators were correlated to blood pressure through the analysis of experimental data.In order to verify the role of the PPG indicators proposed in this study in improving the blood pressure estimation precision,we used data from healthy adults and intensive care unit(ICU)patients the validate the role of the proposed PPG indicators in blood pressure estimation,respectively.The performances of the blood pressure estimation methods were compared between using the PPG indicators proposed in this study and using those in previous studies.First,the effect of 19 new PPG indicators on the accuracy of blood pressure estimation was verified by healthy subjects.22 healthy subjects were recruited,PPG.ECG and noninvasive continuous blood pressure wave were collected simultaneously when they performed mental arithmetic stress and Valsalva’s manoeuvre tasks that could induce BP fluctuations.In order to explore the effect of the PPG indicators proposed in this study on blood pressure estimation,BP estimation models were constructed by means of linear regression using least squares for each indicator.In order to examine if the combination of PPG indicators with commonly used PTT indicator could improve BP estimation accuracy,each of the newly proposed 19 PPG indicators was combined with PTT indicator and then these hybrid indicators were used to develop the(PPG+PTT)models for continuous BP estimation,respectively.The results showed that the best PPG-based BP estimation model could achieve a decrease of 0.31±0.08 mmHg in systolic BP(SBP)and 0.33±0.01 mmHg in diastolic BP(DBP)on estimation errors of grand absolute mean(GAM)and standard deviation(GSD)in comparison to the previously reported PPG-based methods.And the best estimation model based on the combination of PPG and PPT could achieve a decrease(GAM&GSD)of 0.81±0.95 mmHg in SBP and 0.75±0.54 mmHg in DBP in comparison to the PPT-based methodsAfterwards,data from ICU patients were used to further verify the effect of more PPG indicators(45)proposed in this study on blood pressure estimation.As more and more PPG indicators were shown to be promising for continuous blood pressure estimation,simultaneous use of them for blood pressure estimation is not a good strategy.The PPG was first preprocessed for denoising and a signal quality assessment method was used evaluated and eliminate the noisy segment with too much interference.Then,65 PPG features including 45 proposed in this study and 20 reported in previous studies were extracted from the original,the first derivative,and the second derivative wave of PPG.After that,personalized optimal subset of features was selected using a combination of filter and wrapper machine learning method,and multiple linear regression method was used to construct personalized BP estimation models based on the optimal subset of features.The performance of the BP estimation models was evaluated using PPG recordings from 109 ICU patients.The features’ importance and stability were evaluated.The results showed that the proposed PPG-based BP estimation algorithm could decrease from 5.50±7.02 mmHg,5.25±6.74 mmHg to 4.59±6.00 mmHg in systolic BP(SBP)and from 2.90±3.80 mmHg,2.82±3.70 mmHg to 2.47±3.30 mmHg in diastolic(DBP)on estimation errors of mean absolute error(MAE)and standard deviation error(SDE)in comparison to two recently reported methods that used a combination of PTT and PPG features.As more multi-order derivative,multivariate PPG features were included,the accuracy of the BP measurement could be highly improvedIn addition,in practical applications,the PPG recordings are easily corrupted by different interferences.Sometimes,it is very hard to eliminate the noise by commonly used filtering methods.In this study,we proposed a filtering method based on the characteristics of PPG recordings to remove the noisy outliers.Firstly,five characteristics,short-term energy(SE),ascending intensity difference(AID),descending intensity difference(DID),ascending time difference(ATD),and descending time difference(DTD),were chosen as metrics and calculated from cardiac PPG wave.Then the median lines of the five metrics were obtained using a median filter,respectively.An acceptable value range around the median line of each metric was set and used to examine PPG recordings cardiac-circle-by-circle For each cardiac circle,when one or more of its five characteristic values exceed(s)the acceptable range,the PPG recording segment was discarded from further analysis.With this proposed method,the noisy outliers could be efficiently identified from the PPG recordingsMoreover,in this study,we also investigated the relationship between blood pressure variability(BPV)/heart rate variability(HRV)indicators and arterial stiffness under specific stress conditions(cold pressor test)to explore the clinical application and significance of using continuous blood pressure measurement Pulse wave velocity(a measure of arterial stiffness)and continuous blood pressure and heart rate before,during and after the cold stimuli were analyzed in 85 young subjects.The results showed that indicators including diastolic blood pressure average real variability(DBPV_ARV,r=0.25,P<0.05),diastolic blood pressure coefficient of variation(DBPV_SV,r=0.22,P<0.042),standard deviation of R-R interval(HRV_SDNN,r=0.24,P<0.05)and the root mean square successive difference(HRV_rMSSD,r=0.23,P<0.05)were significantly correlated with pulse wave velocity in the cold stimulus phase.However,such correlations were not significant in the baseline phase.This suggests that the degree of arterial stiffness would affect the cardiovascular autonomic nervous system(ANS)function,which could be measured using BPV and HRV indicators,under cold stimulation condition.The higher the degree of arterial stiffness,the greater the pulse wave velocity,and the greater the effect of cold stimulation on blood pressure variability and heart rate variability indicators.Therefore,arterial stiffness can be indirectly evaluated by the response of the blood pressure variability and heart rate variability indicators to external stimuli,which can provide another convenient indirect measurement method for the evaluation of arterial stiffness and play an important role in the prediction of cardiovascular diseasesIn summary,this study focused on cuff-less and continuous blood pressure estimation and application.It is of great significance for the prevention,diagnosis and prognosis of cardiovascular diseases,and has a broad application prospect in the field of mobile health and family medical care services.