Multivariate modelling of the pharmaceutical two-step process of wet granulation and tableting with multiblock partial least squares
✍ Scribed by Johan A. Westerhuis; Pierre M. J. Coenegracht
- Publisher
- John Wiley and Sons
- Year
- 1997
- Tongue
- English
- Weight
- 133 KB
- Volume
- 11
- Category
- Article
- ISSN
- 0886-9383
No coin nor oath required. For personal study only.
✦ Synopsis
The pharmaceutical process of wet granulation and tableting is described as a two-step process. Besides the process variables of both steps and the composition variables of the powder mixture, the physical properties of the intermediate granules are also used to model the crushing strength and disintegration time of pharmaceutical tablets. Multiblock partial least squares (MBPLS) regression is used to model the two-step process. With MBPLS the highly collinear granulate properties can be segregated from the process and composition variables to study separately the influence of both groups of descriptor variables on the tablet properties. This improves the interpretability of an ordinary PLS model. Two different approaches of the MBPLS algorithm are compared for the modelling of the two-step process. One approach suffers severely from correlation between the two descriptor blocks. When the correlation is removed, the approach improves.