Central to the QbD approach is the notion of a Design Space comprised of the "multidimensional combination and interaction of input variables (e.g., material attributes) and process parameters that have been demonstrated to provide assurance of quality." However, limited prescriptive information is available regarding to how to construct such a Design Space and to demonstrate that operation within it "…provides assurance of quality."

Development and calibration of a Design Space typically involves construction of multiple predictive response surface models corresponding to critical drug product attributes.  Such a Design Space is constraint by requirements of meeting multiple response criteria.  Optimizations are typically approached using overlapping mean response or by a desirability function.  However, these approaches fail to account for the uncertainty in model parameters and the correlation structure of the data, which can lead to dramatically misguided conclusions. As shown by Peterson (Journal of Biopharmaceutical Statistics, 18, 959-975, 2008), a Bayesian approach employing posterior predictive distributions addresses both of these limitations.

At the useR! 2009 conference, Blue Reference will present several case studies illustrating the use of Inference for R in conjunction with an assortment of R packages towards the construction of design spaces for representative pharmaceutical products using Bayesian Methods.  An abstract of the presentation can be obtained at its linked title below:

Bayesian Approach to the Specification of Design Space in Quality by Design

Also to be presented at the useR! 2009 conference, is a paper illustrating the use of Inference to construct dynamic applications, as in Inference for QbD. An abstract of this presentation can be obtained at its linked title below:

Microsoft Office Dynamic Documents as R Applications