|
Activity Number:
|
345
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #306669 |
|
Title:
|
Design of Experiments for Parameter Estimation in Compartmental Models
|
|
Author(s):
|
Michael DeVasher*+ and J. Michael Hardin
|
|
Companies:
|
The University of Alabama and The University of Alabama
|
|
Address:
|
P.O. Box 864676, Tuscaloosa, AL, 35486,
|
|
Keywords:
|
compartmental models ; nonlinear design ; pharmacokinetics
|
|
Abstract:
|
The design of experiments to estimate the parameters of a nonlinear model requires prior knowledge of these same unknown parameters. As such, an optimum experimental design hinges on the ability of the researcher to incorporate previously gained expertise into the new research. Little work has been done in the design of experiments aimed at estimating the parameters of a compartmental model, especially incorporating prior information on the parameters. Through the use of smoothing splines and locally weighted regression, we construct designs using more informative prior distributions that provide for a decrease in the variability of the estimated parameters. An analysis of the improved efficiency of such designs is discussed. Also, the resistance of the proposed designs to various inaccuracies in the prior distributions is explored.
|
