Studies On Quality Control Of Herbal Medicine In Yiqing Capsule Based On Near-infrared Spectroscopy

Traditional Chinese medicine(TCM)has played an important role in the prevention,treatment and recovery of major diseases and has been widely recognized at home and abroad because of its unique concept of treatment and health..However,the source of TCM is complex,and there are obvious differences in quality due to factors such as growth conditions,harvest seasons,processing and transportation approaches.The quality control of TCM has become the focus of attention all over the world.In recent years,near-infrared spectroscopy(NIRs)technology has provided technical support for the rapid analysis of TCM.Its high efficiency and non-destructiveness show great potential in the field of TCM analysis.In this study,the reaseach object is the herbal medicine of Yiqing capsule including Scutellariae radix,Rhei radix et Rhizoma and Coptidis rhizome.NIRs combined with chemometrics was applied to achieve rapid detection and quality control of three TCM,as well as the monitoring and prediction of baicalin content in the extraction and concentration process.The main research contents and achievements are as follows:(1)Taking baicale content in Scutellariae radix,extraction process and concentration process as research objects,NIRs technology was applied to establish a rapid quantitative model.A total of 79 different batches of Scutellariae radix were collected and the production process of Yiqing capsules was simulated in the lab.The spectra of three stages were collected and characteristic variable selection algorithm was applied to extract the characteristic wavenumbers from the raw spectra.The processed spectral variables are reduced by more than 95%,effectively eliminating redundant variables,and significantly reducing the complexity of the model.The modeling efficiency is improved.The linear algorithm partial least squares regression(PLSR)and non-linear algorithm extreme learning machine methods were applied to construct the model and the model performance were compared.Finally,the best calbration models for herb,extraction process and concentration process are constructed.The relative standard errors of prediction set(RSEP)are all less than 9%,the rapid detection of baicalin content in different processes is realized.(2)Based on NIRs technology,rapid determination of loss on drying,extracts,free anthraquinone and total anthraquinone in Rhei radix et Rhizoma was carried out.A total of 142 samples from different batches were collected for near-infrared spectroscopy collection and content determination.After spectral collection,the PLSR method and particle swarm optimization-least squares-support vector machines were used to construct quantitative model.The best model was selected according to the characteristics of the substance.In addition,the built models were used for real-time release testing,and the qualified samples and abnormal samples were accurately identified by constructed models.Finally,based on the ?-content tolerance interval,a comprehensive verification and evaluation of the free anthraquinone content and total anthraquinone content models was performed,the relative deviation,accuracy,and accuracy were included.The model validation could ensure the accurate release of Rhei radix et Rhizoma.(3)Combined with NIRs and global model,the rapid analysis of Coptidis Rhizoma(CR)and Phellodendri Chinensis Cortex(PC)was explored.Firstly,the rapid detection model of CR and PC was established using the PLSR method.At the same time,based on the same composition and similar clinical application of CR and PC,the global model of two TCM was established in this study for rapid prediction of moisture and berberine content.The experimental results showed that the global model could achieve rapid quantitative analysis of CR and PC,the RSEP values were all less than 5.5%.The rich sample set and broadened prediction range make the model more accurate.Especially for the prediction of the moisture content in the CR and the berberine content in PC,the global model obtained higher prediction accuracy than the single species model.The global mode provides new ideas for the quality control of CR and PC.