Estimation of design space for an extrusion–spheronization process using response surface methodology and artificial neural network modelling

Abstract

The application of the Quality by Design principles is one of the key issues of the recent pharmaceutical developments. In the past decade a lot of knowledge was collected about the practical realization of the concept, but there are still a lot of unanswered questions. The key requirement of the concept is the mathematical description of the effect of the critical factors and their interactions on the critical quality attributes (CQAs) of the product. The process design space (PDS) is usually determined by the use of design of experiment (DoE) based response surface methodologies (RSM), but inaccuracies in the applied polynomial models often resulted in the over/underestimation of the real trends and changes making the calculations uncertain, especially in the edge regions of the PDS. The completion of RSM with artificial neural network (ANN) based models is therefore a commonly used method to reduce the uncertainties. Nevertheless, since the different researches are focusing on the use of a given DoE, there is lack of comparative studies on different experimental layouts. Therefore, the aim of present study was to investigate the effect of the different DoE layouts (2 level full factorial, Central Composite, Box-Behnken, 3 level fractional and 3 level full factorial design) on the model predictability and to compare model sensitivities according to the organization of the experimental data set. It was revealed that the size of the design space could differ more than 40% calculated with different polynomial models, which was associated with a considerable shift in its position when higher level layouts were applied. The shift was more considerable when the calculation was based on RSM. The model predictability was also better with ANN based models. Nevertheless, both modelling methods exhibit considerable sensitivity to the organization of the experimental data set, and the use of design layouts is recommended, where the extreme values factors are more represented.

A two-step approach for fluidized bed granulation in pharmaceutical processing: Assessing different models for design and control

“…It is clear that using RSM-based evaluation results in a better correlation with the data than PLS or ANN. This could be due to the fact that PLS and ANN functions may not suitable for modeling a complex surface since they cannot detect slight changes in parameters and considerably under/overestimate the values in those regions where the response function changes its sense [52]. In this study, it has been shown that the nonlinear approach based on RSM is better at generalization and prediction than the multiple linear technique PLS.…”

Detoxification of linseed-sunflower meal co-extrudate: Process prediction

“…Recently, mathematical modelling has been increasingly used for the study of the given systems. Artificial Neural Network (ANN) models have gained momentum for modelling and control of extrusion processes (Li and Bridgwater, 2000;Cubeddu et al, 2014;Sovány et al, 2016). ANN models are recognized as a good modelling tool since they provide the empirical solution to the problems from a set of experimental data, and are capable of handling complex systems with nonlinearities and interactions between decision variables (Shankar et al, 2007;Fan et al, 2013).…”

Aspects to be considered in case of variable surfaces modelling

“…The procedure proposes to triangulate in an optimum way any type of surface through a series of edge flip operations. Those operations should keep and maintain the initial topology of any surface in such a way that will always be a valid mesh which would allow re-triangulations of the same surface as the process advances [2].…”