University of Newcastle upon Tyne

School of Mathematics and Statistics

Statistics Seminars 2005-2006

7 October 2005, L401, 2:00pm

John Matthews

'Bayesian' designs for Michaelis-Menten kinetics

 Abstract

The formal design of experiments usually attempts to arrange observations in such a way as to optimise a criterion which encapsulates the aims of the experiment.  The traditional alphabetic criteria of A- D-, E- optimality are examples of this.  For non-linear models matters are complicated by the fact that these criteria will generally depend on unknown parameters.  An approach which is widely described as Bayesian optimal design can be adopted to overcome.

This methodology will be illustrated by an example from enzymology, where the observations are taken from a process governed by the Michaelis-Menten equation.  Bayesian D-optimal designs will be derived and some seemingly paradoxical properties of the designs noted.  A resolution of these issues will be found by considering the foundations of the ‘Bayesian’ D-optimal designs.

Return to timetable for 2005-06 seminars

School of Mathematics and Statistics Homepage