Activity Number:
|
428
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, August 9, 2006 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Section on Bayesian Statistical Science
|
Abstract - #307149 |
Title:
|
Join-Point Analysis of Survival Data
|
Author(s):
|
Sandra Hurtado Rua*+ and Sanjib Basu
|
Companies:
|
Northern Illinois University and Northern Illinois University
|
Address:
|
Mathematical Sciences, DeKalb, IL, 60115-2888,
|
Keywords:
|
Bayesian analysis ; joinpoint regression ; survival analysis ; hazard function ; cancer data
|
Abstract:
|
The identification of change of tendency over time in cancer data is important for determining agents that affect cancer. Join-point models help to describe continuous changes in trends of data and when such changes occur. Join-point models play an important role since they help to measure the impact of health policies and treatments, and in general, they help to locate and explain changes in the survival curve. We model join-point hazard and survival functions in cancer data that account for censored and uncensored observations. Given k join-points, the posterior distributions of the parameters of their models are estimated using Bayesian analysis with Markov chain Monte Carlo sampling. We further propose model selection approaches to determine the number of join-points in the model. We illustrate the method using monthly data from the SEER cancer database.
|