A near infrared spectroscopy（NIRS） approach was established for quality control of the alcohol precipitation liquid in the manufacture of Codonopsis Radix. By applying NIRS with multivariate analysis, it was possibl...A near infrared spectroscopy（NIRS） approach was established for quality control of the alcohol precipitation liquid in the manufacture of Codonopsis Radix. By applying NIRS with multivariate analysis, it was possible to build variation into the calibration sample set, and the Plackett-Burman design, Box-Behnken design, and a concentrating-diluting method were used to obtain the sample set covered with sufficient fluctuation of process parameters and extended concentration information. NIR data were calibrated to predict the four quality indicators using partial least squares regression（PLSR）. In the four calibration models, the root mean squares errors of prediction（RMSEPs） were 1.22 μg/ml, 10.5 μg/ml, 1.43 μg/ml, and 0.433% for lobetyolin, total flavonoids, pigments, and total solid contents, respectively. The results indicated that multi-components quantification of the alcohol precipitation liquid of Codonopsis Radix could be achieved with an NIRS-based method, which offers a useful tool for real-time release testing（RTRT） of intermediates in the manufacture of Codonopsis Radix.展开更多
基金supported by the National Basic Research Program(973)of China(No.2012CB518405)the Zhejiang Traditional Medical Science and Technology Projects(No.2015ZB023),China
文摘A near infrared spectroscopy（NIRS） approach was established for quality control of the alcohol precipitation liquid in the manufacture of Codonopsis Radix. By applying NIRS with multivariate analysis, it was possible to build variation into the calibration sample set, and the Plackett-Burman design, Box-Behnken design, and a concentrating-diluting method were used to obtain the sample set covered with sufficient fluctuation of process parameters and extended concentration information. NIR data were calibrated to predict the four quality indicators using partial least squares regression（PLSR）. In the four calibration models, the root mean squares errors of prediction（RMSEPs） were 1.22 μg/ml, 10.5 μg/ml, 1.43 μg/ml, and 0.433% for lobetyolin, total flavonoids, pigments, and total solid contents, respectively. The results indicated that multi-components quantification of the alcohol precipitation liquid of Codonopsis Radix could be achieved with an NIRS-based method, which offers a useful tool for real-time release testing（RTRT） of intermediates in the manufacture of Codonopsis Radix.
