期刊
JOURNAL OF PHARMACEUTICAL AND BIOMEDICAL ANALYSIS
卷 145, 期 -, 页码 346-355出版社
ELSEVIER
DOI: 10.1016/j.jpba.2017.06.070
关键词
NIR spectroscopy; Raman spectroscopy; CHO cultivation; Process monitoring; Process Analytical Technology; PAT
资金
- Gedeon Richter's Talentum Foundation (Budapest, Hungary)
In this study, near-infrared (NIR) and Raman spectroscopy were compared in parallel to predict the glucose concentration of Chinese hamster ovary cell cultivations. A shake flask model system was used to quickly generate spectra similar to bioreactor cultivations therefore accelerating the development of a working model prior to actual cultivations. Automated variable selection and several pre-processing methods were tested iteratively during model development using spectra from six shake flask cultivations. The target was to achieve the lowest error of prediction for the glucose concentration in two independent shake flasks. The best model was then used to test the scalability of the two techniques by predicting spectra of a 101 and a 1001 scale bioreactor cultivation. The NIR spectroscopy based model could follow the trend of the glucose concentration but it was not sufficiently accurate for bioreactor monitoring. On the other hand, the Raman spectroscopy based model predicted the concentration of glucose in both cultivation scales sufficiently accurately with an error around 4 mM (0.72 g/l), that is satisfactory for the on-line bioreactor monitoring purposes of the biopharma industry. Therefore, the shake flask model system was proven to be suitable for scalable spectroscopic model development. (C) 2017 Elsevier B.V. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据