Pulse Signal Analysis Method Based On Multi-scale Cross Approximate Entropy And Its Application

Pulse signal is an important physiological signal of the human body,which is closely related to the degree of arteriosclerosis,blood vessels and other health conditions of the human body.Traditional methods have the influence of subjective factors in the way that the physiological state of the human body is assessed based on pulse.In terms of the analysis of the signal,the previous research was carried out from the perspective of the time domain and the frequency domain.In view of the non-stationary and nonlinear characteristics of the pulse signal,in this thesis we propose to use a nonlinear dynamics method—Multiscale Cross-Approximate Entropy(MCAE)to analyze and process it.The proposed method is applied to the differential analysis study of pulse signals between healthy people and diabetic patients.The specific research contents are as follows:First,the human pulse signal is acquired in a noninvasive manner using photoplethysmography waves.Because the acquired signal has problems such as unobvious feature points,preprocessing is performed in Ensemble Epithelial Mode Decomposition(EEMD)manner.And white noise was added before signal decomposition to avoid the phenomenon of modal aliasing.EEMD performs better than the pretreatment of Empirical Mode Decomposition(EMD),effectively removes the interfering part and retains the signal characteristics.Secondly,to locate the peak and valley points during the pulse signal cycle and extract the Crest time(CT),Peak-to-Peak Interval(PPI)sequence data.The sample entropy was calculated for the pulse signal CT and PPI sequences of 81 healthy people and diabetic people.The results showed that the entropy values of diabetic patients were 0.1 lower than that of healthy people.Calculating the multiscale entropy after coarsening the data revealed that the values of multiscale entropy differed by 0.2 between the two groups,and the method in this paper discriminated better.Finally,the MCAE algorithm in this study was used to analyze the entropy value of pulse signals in MCAE3 and MCAE4 under different physiological states,showed a significant difference of 0.33(p<0.001).Meanwhile,the instantaneous frequency index of pulse signal was calculated for comparison.The results showed that the instantaneous frequency value of diabetic patients was increased by 0.25 compared with that of healthy people,and the difference was not as obvious as that of MCAE3 and MCAE4.In correlation analysis,MCAE3 was significantly correlated with Hb A1c(p<0.001)and negatively correlated(r =-0.399)to a higher extent than the instantaneous frequency value(r = 0.227).MCAE3 exemplifies the pulse signal variability in different physiological states in the form of computational sequence complexity.Through the above research,this thesis realizes the acquisition of human pulse signal in a non-invasive way,pre-processing by EEMD,the correlation between the obtained waveform and the original signal is 0.979,which ensures that the two synchronized time series of CT and PPI are accurately extracted.The difference in entropy between the two groups of people at MCAE 3 and 4 is 0.33,showing a significant difference(p<0.001),and there is a significant negative correlation with human blood sugar content(r=-0.399).Multiple linear regression analysis shows that the HbA1c coefficient in the entropy value is0.093.Compared with the coefficient of 0.028 in the instantaneous frequency,the proportion of HbA1c has a greater influence.This study provides important reference value for the application of entropy theory to analyze human physiological signals.