Optimal Experimental Designs for Pharmacokinetic Models

In this book we provide extensions to experimental designs based on variations of the Michaelis-Menten model used in the study of pharmacokinetic processes. New optimal designs have been found for the modification of the Michaelis-Menten model consisting of adding a linear term to the initial model. In the homoscedastic and heteroscedastic cases are obtained the D-optimal designs. The concept of T-optimality is used to deal with the problem choosing a model before having the data, as opposed to after performing a statistical hypothesis test. We obtain T-optimal designs for discriminating between two homoscedastic models with normal distributions. We have found, by extrapolation, a new criterion for discriminating between two non-normal models, which we have called the KL-criterion, seeing as it is defined in terms of the Kullback-Leibler distance. It is applied to discriminate between a Michaelis-Menten model and its modifications, where the former has a log-normal distribution and the latter a gamma distribution. The KL-criterion generalizes the previous criteria found in the literature for multi-response heteroscedastic models as well as for binary and generalized models.

Autorentext

Paula Camelia Trandafir earned her Ph.D. from the University of Castille and La Mancha. Her field of expertise is optimal experimental design, in which she is the author of various research papers and a participant in international conferences. Dr. Trandafir is currently a professor at the Public University of Navarre.