|
Activity Number:
|
269
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Tuesday, August 8, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #305553 |
|
Title:
|
Bayesian Inference for NLME Models Involving ODEs
|
|
Author(s):
|
Sujit Ghosh*+ and Lovely Goyal
|
|
Companies:
|
North Carolina State University and North Carolina State University
|
|
Address:
|
2501 Founders Drive, Raleigh, NC, 27695,
|
|
Keywords:
|
Bayesian ; mixed effects ; nonlinear ; MCMC ; PKPD ; ODE
|
|
Abstract:
|
Nonlinear mixed effects (NLME) models, such as PK/PD and PBPK, have been used widely in various biomedical fields. Such models are desirable because they allow us to study the dynamics of a drug within an individual represented by a set of ordinary differential equations (ODEs). Likelihood-based inference becomes challenging and computationally intensive when the system of ODEs is large and no analytical solution is available. We propose a new approach based on Euler approximation that allows us to obtain a tractable likelihood that approximates the original likelihood as the grid size approaches zero at a certain rate. We also develop efficient MCMC methods within a Bayesian framework to obtain parameter estimates. To illustrate our method, we apply it to data on HIV patients and present simulation studies for model validation and comparison to other competing approaches.
|
