ABSTRACT
Near-infrared spectroscopy was used to monitor the end point of the elution process for the purification of Gardenia jasminoides Ellis (gardenia) extract. A partial least square model was built to quantify the concentration of geniposide. A new strategy was coupled with multivariate statistical process control to update the original multivariate calibration model to correct for fluctuations among batch processes and to enhance the robustness of the model. After updating the model twice by the new strategy, the root mean squared error of prediction decreased from 0.505 to 0.350 mg mL−1, and the interval between the end points determined by near-infrared spectroscopy and high-performance liquid chromatography was reduced from 15 to 3 min. Compared with other model updating methods, this strategy significantly improved the prediction capabilities of the near-infrared calibration model with smaller prediction errors while requiring fewer samples in the calibration set.
Acknowledgments
The authors would like to acknowledge the help of the Joint Development Program supported by Beijing Municipal Education Commission—Key Laboratory Construction Project (Study on the Integrated Modeling and Optimization Technology of the Chained Pharmaceutical Process of Chinese Medicine Products).