|
Activity Number:
|
30
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 6, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #306065 |
|
Title:
|
A Bayesian Multivariate PK/PD Model for Analyzing Cortisol Circadian Rhythm in a Depression Study
|
|
Author(s):
|
Niko Kaciroti*+ and Trivellore E. Raghunathan and Delia Vazquez
|
|
Companies:
|
University of Michigan and University of Michigan and University of Michigan
|
|
Address:
|
300N Ingalls Building, Ann Arbor, MI, 48109,
|
|
Keywords:
|
Markov chain Monte Carlo ; nonlinear hierarchical models
|
|
Abstract:
|
This paper presents results from a study of depression among new mothers and their infants using their cortisol circadian rhythm (CCR). We construct a multivariate Pharmacokinetic-Pharmacodynamic (PK/PD) model to jointly estimate the CCR for mother and infant as well as the strength of the interdependence between their individual CCR. A non-linear random effects model is used where each individual trajectory has its own parameters. We assume that the subject-specific parameters are normally distributed around the parameters of the overall population-average trajectory. The model is fitted using the Bayesian approach implemented through MCMC. The proposed model is implemented to asses the relationship between alteration of the mother's CCR and depression. Furthermore to assess whether infants whose CCR is strongly related to their mother's CCR are likely to show better neurodevelopment.
|
