Study On Fast Detection Method Of Low API-signal Drugs By Raman Spectroscopy

The endless emergence of counterfeit and inferior medicines in the drug distribution market is a huge threat to the health of consumers and the safety of global public health system.Drug fast detection technique has emerged as the times requirement.It has been widely used in the detection of counterfeit drugs with its advantages of rapidness,accuracy,convenience,and economy.Among them,the spectral characteristics of normal Raman spectroscopy(NRS)are abundant,and it is one of the important analysis tools for fast drug detection because of its advantages of rapid,non-destructive,simple pretreatment or no pretreatment.However,due to the low content of active pharmaceutical ingredient(API)in some chemical drugs,or the weak Raman response to its API,these types of drugs(low API-signal drugs,LASIDs)that exhibit the Raman spectra with low API signal.Their Raman spectra closely resemble the Raman spectrum of a certain excipient,and the spectral information of the API is masked by the spectral information of the excipient.Combined with general chemometric methods,Raman spectroscopy analysis technology is difficult to achieve rapid inspection of such drugs.In order to achieve rapid detection of Raman spectroscopy for LASIDs,this thesis begins a series of discussions and studies of the determination of LASIDs,the Raman spectral characteristic of LASIDs,scalable moving window,scalable moving-window similarity,weighted scalable moving-window similarity(WSMS),logistic regression(LR),Bayesian discrimination and partial least square and so on.Finally,two methods are proposed for the fast detection of LASIDs by Raman spectroscopy.One is the scalable moving-window similarity and Bayesian discriminant algorithm,and the other is weighted scalable movingwindow similarity and logistic regression algorithm.The results showed that both of these methods can achieve the rapid detection of LASIDs by Raman spectroscopy,which is highly efficient and convenient.In this thesis,firstly,the related introduction of Raman spectroscopy analysis technique,LASIDs,chemometrics theory and its progress in pharmaceutical research are summarized.Then it elaborates the research and application of Raman spectroscopy combined with chemometrics method in the detection of LASIDs.Part ?.Overview.Raman spectroscopy analysis technique for drug detection is efficient and convenient.It has been widely used in the field of drug rapid inspection.However,in the Raman spectra of LASIDs,the spectral information of its API is basically obscured by the spectral information of its excipient.The general Raman spectroscopy method is difficult to achieve fast detection of such drugs.The scalable moving-window similarity and weighted scalable moving-window similarity can extract the spectrum information of its API in the Raman spectrum of LASIDs.According the spectral information of given drugs,the chemometric methods such as Bayesian discrimination,logistic regression and partial least square can be applied to construct analytical models to achieve drug fast detection.In this thesis,these methods are improved and applied to rapidly detect LASIDs combine with Raman spectroscopy,which provide a new method for LASIDs' fast detection.Part?.Research on fast detection of LASIDs based on scalable moving-window similarity and Bayesian discrimination.This study proposes scalable moving-window similarity and Bayesian discriminant method to fast distinguish the authenticity of LASIDs by Raman spectroscopy.First of all,the window size of scalable moving-window is dynamically adjusted according the peak width of API spectrum.Within the window,similarity between the LASID spectrum and the API spectrum,and similarity between the LASID spectrum and the excipient spectrum are calculated.Then,windows that highlights contribution of the API spectral signal to the LASID spectral signal are chosen.These chosen windows are as variables to construct Bayesian discriminant model for identifying LASIDs' authenticity.The results showed that the accuracy rate of discrimination model is 94.7% for the identification of LASIDs in training set,and accuracy rate of 95.6% for the identification of LASIDs in testing set.The Bayesian discriminant model constructed by scalable movingwindow similarity and Bayesian discrimination can quickly distinguish the authenticity of LASIDs.Part ?.Research on fast detection of LASIDs based on weighted scalable movingwindow similarity and logistic regression.This study proposes weighted scalable movingwindow similarity and logistic regression for rapid detection of the LASIDs.Firstly,weighted scalable moving-window similarity is used to analyze the Raman spectrum information of LASIDs and its API and related excipients.After weighting the scalable moving-window similarity,the contribution of API spectral signal to the spectrum signal of LASID is highlighted.At the same time,the interference of the excipient spectrum signal is reduced.Then the API spectral peak signal masked by excipient spectral information is found out from the LASID spectrum.Then,a logistic regression algorithm is used to construct a prediction model for the analysis of LASIDs' authenticity.The results show that the goodness of fit test for this model have a P value of 0.855,a sensitivity of 90.6%,and a specificity of 94.0%.The prediction model constructed based on the weighted scalable moving-window similarity and the logistic regression algorithm can quickly detect LASIDs.As there are drugs other than LASIDs in the drug distribution market,there are other types of drugs,such as high API content drugs(the proportion of API content in the drug is relatively high,about 40% to 80% w/w),and drugs that have a stronger Raman response to the API(high API-signal drugs,HASIDs).In order to better apply the weighted scalable moving-window similarity and logistic regression algorithm to the drug fast detection,the study uses the prediction model constructed by this algorithm for rapidly detecting the real LASIDs and other 3 types of drugs.The results showed that the prediction model can accurately detect LASIDs and other 3 types of drugs with 100% accuracy.The weighted scalable moving-window similarity and logistic regression algorithm combined with Raman spectroscopy can rapidly detect drugs,which provides a new method and injects new vitality for rapid drug detection.