|
Activity Number:
|
193
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 4, 2008 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #302287 |
|
Title:
|
Effects of Missing and Censored Data for Nonlinear Models Involving ODEs
|
|
Author(s):
|
Sujit Ghosh and Haojun Ouyang*+
|
|
Companies:
|
North Carolina State University and North Carolina State University
|
|
Address:
|
Bioinformatics Program, Raleigh, NC, 27695-7566,
|
|
Keywords:
|
Bayesian inference ; Censored data ; Data augmentation ; Missing data ; Monte Carlo simulation
|
|
Abstract:
|
The Bayesian Euler's Approximation Method (BEAM) has recently been proposed to estimate the parameters in a non-linear model involving with ODEs, especially when analytical closed form solutions are not available. In this article, the BEAM is extended to handle datasets with missing or censored observations. The proposed method is based on data augmentation algorithm. A simulation study based on growth colonies of paramecium aurelium is presented to compare the performances of the proposed method for various percentages of missing and censored data cases and results are compared to complete data case. Finally the method is illustrated with a real data of AIDS Clinical Trials Group Protocol 315.
|
