Several Technologies To Improve The Accuracy Of Noninvasive Detection Of Blood Components By Dynamic Spectroscopy

The contents of blood components in human are key indexes to reflect health condition and play an irreplaceable role in the prevention and diagnosis of diseases.The traditional invasive biochemical detection not only has complicated procedures and pollution,but also causes pain to the subjects,furthermore it can't achieve long-term monitoring in real time.Therefore,non-invasive detection of blood components has become the research hotspots in recent years.Among these non-invasive detection methods,dynamic spectroscopy has obvious advantages and broader development prospects because it can theoretically eliminate the influence of individual differences and changes of measurement conditions.But there still exist some unsolved problems in dynamic spectroscopy.In this paper,research has been made in combined effects between PPG preprocess and dynamic spectrum extraction and nonlinear correction by grouping modeling.The non-invasive detection accuracy of blood components based on dynamic spectrum not only depends on the accuracy of each step individually,but also the proper cooperation between these steps.Therefore,combined effects between PPG preprocess and dynamic spectrum extraction were investigated: dynamic spectrum extraction method itself has some but insufficient ability of noise suppression,and therefore PPG preprocess method should be chosen properly to complement dynamic spectrum extraction method's deficiency in noise suppression and further improve the extraction accuracy of dynamic spectrum.To examine the combined effects between PPG preprocess and dynamic spectrum extraction,76 clinical samples were used as subjects investigated,three filters(namely zero phase-shift bandpass filter,wavelet filter and empirical mode decomposition filter)were applied for PPG preprocess and single-trial estimation was applied for dynamic spectrum extraction.Partial least square regression was applied for model establishment between dynamic spectrum and hemoglobin content.The results showed that in comparison with no preprocess,the root mean square error of prediction set is reduced from 14.1003 g/L to 10.7270 g/L,11.1018 g/L,11.2768 g/L respectively with zero phase-shift bandpass filter,wavelet filter,EMD filter.Zero phase-shift bandpass filter not only keep the phase of original PPG signals unchanged,but also complements the deficiency of single-trial estimation in suppressing high frequency noises.As a result,zero phase-shift bandpass filter is more effective to improve robustness of the model and prediction accuracy of hemoglobin content.Through considering the combined effects between steps,non-invasive detection accuracy of blood components was further improved and the pace of dynamic spectroscopy into clinical application was accelerated.It also has referential significance to quantitative analysis with spectroscopy.To correct the non-linearity caused by light scattering in non-invasive detection of hemoglobin content based on dynamic spectrum,a new modeling analysis method was proposed: grouping modeling according to the content of analyte.275 clinical samples were used as subjects investigated and partial least square regression was applied for model establishment.Prediction performance comparison was made between the proposed method and non-grouping modeling method.Experimental results show that the root mean square error of the prediction set by the proposed method was reduced by about 10%.The results demonstrated that the proposed method could linearize the model to a certain extent and reduce the effects of non-linearity on hemoglobin detection.This research has great importance for improving detection accuracy of blood components based on dynamic spectroscopy and provides a new thinking for correcting non-linearity in the qualitative analysis of other analytes by spectroscopy.