Optimal Bayesian Design of Experiments Applied to Nitroxide-Mediated Radical Polymerization

Bayesian design of experiments is a powerful method which offers several distinct benefits over standard experimental designs. The basics of the method are briefly described, followed by four case studies giving a step-by-step illustration of its application to both bimolecular and uni-molecular NMRP. Firstly, the Bayesian design is an improvement with respect to information content retrieved from process data. It allows one to change the levels of factors with relative ease and is flexible and "cost"-effective with respect to the number of experiments. More importantly, the method has the ability to incorporate into the design prior knowledge coming from a variety of sources. Diagnostic criteria can shed more light on the quality of prior knowledge and the significance of estimated effects.