The Application Of Near Infrared Spectroscopy For Quality Control Of Codonopsis Pilosula And Fu Fang Dan Shen Tablets

Nowadays, quality control of traditional Chinese medicines (TCMs) have been paid more and more attention. With the development of analytical technology and the efforts that analysts has been made, the quality control of TCMs is gradually moving towards an integrative and comprehensive direction. Currently, developing a feasible way of comprehensive rapid and efficient quality control of TCMs has become an important task. Among many comprehensive methods, near infrared spectroscopy (NIRs) analysis has its unique advantages in the rapid analysis of TCMs. Therefore, the aim of this study was to develop NIRs analysis methods for comprehensive, rapid and efficient quality control of Codonopsis pilosula and Fu Fang Dan Shen tablets (FFDST).In Chapter 1 and Chapter 2, some relevant contents were reviewed, including status of quality control of TCMs, methods of quality control, NIRs analysis technology, the background and foundation of this study, et al.In chapter 3 of this article, Codonopsis pilosula from different regions and species were surveyed by NIRs. In the first work. Random forests (RF) was applied to build the classification models of Codonopsis pilosula from different species according to original and pretreated NIR spectrum, and the classification results were compared. The results indicated that the proposed classification models, whose spectrum was pretreated by standard normal variate transformation (SNV) together with derivation, showed a better performance. The classification accuracy of training set and test set was 100% and 93.75%, respectively. In the second work, the original spectrum was processed by SNV together with derivation. Then RF, Decision tree (DT) and k nearest neighbor (KNN) were used to build the classification models of Codonopsis pilosula from different regions and the classification results were compared. The best model was proposed by RF and the classification accuracy of training set and test set was 95.38% and 100%. The results obtained from this study demonstrated that the proposed model based on NIR spectrum could discriminate the Codonopsis pilosula from different origin and species, and these methods could be used for comprehensively and rapidly control the quality of Codonopsis pilosula.In Chapter 4, Mice tail bleeding time was measured to estimate the efficacy of FFDST. The quantitative spectrum-activity relationship (QSAR) between pretreated NIR and activating blood circulation activity of FFDST was established using partial least square regression (PLSR). The predictive ability of the proposed QSAR model was internally validated. In the proposed QSAR model, the best model was the one whose spectrum was pretreated by SNV combined with derivation. The number of best principal factor of this model was 3, the value of the R2 was 0.666, RMSELoo was 0.046, RMSEtr was 0.022. The results showed that the proposed model has the better predictive ability internally and demonstrated simultaneously that the QSAR model could provide a new strategy for comprehensive, rapid and efficient quality control of TCMs.