|
Activity Number:
|
388
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 9, 2006 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #307374 |
|
Title:
|
Local Likelihood Models for Disease Cluster Modeling: a Space-Time Extension
|
|
Author(s):
|
Monir Hossain*+ and Andrew B. Lawson
|
|
Companies:
|
University of South Carolina and University of South Carolina
|
|
Address:
|
800 Sumter Street, Columbia, SC, 29208,
|
|
Keywords:
|
local likelihood ; space-time modeling ; random effects ; Bayesian hierarchical modeling
|
|
Abstract:
|
Data-dependent clustering (Lawson 2006 and Hossain and Lawson 2005) assumed that the relative risk of a disease is defined to be a function of the `local' concentration of cases. This is the idea of local likelihood and the model based on local likelihood has found a novel application to disease cluster modeling. This paper extends the spatial local likelihood model (Hossain and Lawson, 2005) to spatio-temporal settings. The approach is based on extending the circler window for the lasso parameter to a cylindrical window where the base representing space and height representing time. We employ a Bayesian hierarchical modeling approach with a joint implementation of Gibbs and Metropolis-Hasting MCMC computational methods to obtain posterior estimates of all model parameters. We use the lung cancer mortality in 88 counties of Ohio State for the year 1968-88 to illustrate the method.
|
