Design And Research Of Unconstrained Sleep Health Monitoring System Based On BCG

The quality of sleep affects people's mental outlook.Generally,polysomnography(PSG)is used to evaluate sleep quality by monitoring various parameters in clinical practice,such as electroencephalogrphy(EEG),electrocardiogram(ECG),electrooculogram(EOG),electromyogram(EMG),photoplethysmography(PPG),blood oxygen saturation,respiratory rate and so on.People need to wear a variety of sensors which closely contact with the human body by using PSG.However,if people wear so many sensors before going to bed every night in home life,to a certain extent,it has caused an impact on the quality of sleep.This thesis provides a kind of sleep health monitoring system which is unconstrained and does not disturb sleep based on the guarantee of validity and accuracy.In this method,the polyvinylidene fluoride sensor(PVDF)is placed under the sleeping pad to collect the weak signal of the periodic pressure of the body caused by pumping blood from the heart of the human body,namely ballistocardiogram(BCG).Three health indicators of sleeping can be extracted from the BCG signals: the heart rate,the respiration and the movement of the human body.Long-term monitoring of apnea and night out of bed is achieved through respiratory and body movement indicators,and the classification of sleep stages is completed through multiple indicators of heart rate variability(HRV).This thesis implemented the design of ECG signals acquisition circuit and the design of BCG signal acquisition circuit through the analysis of various links in the analog weak signal acquisition process from the hardware aspect.Then the preprocessing process of ECG signals and cardiac shock signals is completed in the upper computer,including baseline drift removal,artifact removal,and least mean square adaptive digital filtering.Through the extraction of R wave from ECG and J wave from BCG signal,the calculation of HRV in time domain,frequency domainand nonlinear analysis is completed.Then,two experiments are used to verify this system,firstly,10 subjects' ECG signals and BCG signals were collected synchronously in the experiment,and the paired T-test was used to verify that there was no significant difference between the two signals in the analysis of HRV.It was proved that the BCG signals can replace the ECG signals to complete the traditional calculation of the indicators of HRV;During the experiment,the subjects simulated sleep apnea and bed leaving at night,and through the breathing and body movement signals extracted from this system,this system can effectively monitored sleep apnea and night out of bed.Secondly,the ECG signals in Sleep Heart Health Study(SHHS)PSG Database sleep monitoring database is used to model and test,and the related index of heart rate variability is extracted as the feature input,and the classification of sleep stages is completed by using radial basis function(RBF)in support vector machine(SVM).Among them,the accuracy rate of three classification(WAKE,NREM,REM)reached 70.26%,and the accuracy rate of five classification(WAKE,REM,NI,NII,NIII)reached 63.49%.The classification results prove to a certain extent that the plan for sleep staging through heart rate variability is feasible.The experimental results laid the foundation for undisturbed sleep monitoring in a home environment.Compared with other non-contact monitoring solutions,this method has the advantages of strong anti-interference ability,low cost,and no interference with normal sleep.Compared with similar domestic products,the hardware circuit designed by the system can better collect the BCG signals,and to a certain extent,the accuracy of the calculation of sleep indicators such as heart rate is guaranteed.