|
Activity Number:
|
535
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 10, 2006 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #306305 |
|
Title:
|
Empirical Bayes Estimation for Additive Hazards Regression Models
|
|
Author(s):
|
M. Brent McHenry*+ and Debajyoti Sinha and Stuart Lipsitz and Malay Ghosh
|
|
Companies:
|
Bristol-Myers Squibb Company and Medical University of South Carolina and Brigham and Women's Hospital and University of Florida
|
|
Address:
|
5 Research Parkway, Wallingford, CT, 06492,
|
|
Keywords:
|
gamma process ; marginal likelihood ; posterior process
|
|
Abstract:
|
We develop a novel empirical Bayesian framework for the semiparametric additive hazards regression model of Aalen (1980) by using a gamma-process prior on the unknown baseline cumulative hazard. The marginal likelihood obtained via integrating the prior process can be maximized using standard statistical software and the empirical Bayes estimates of regression parameters, survival curves and their corresponding standard errors have easy to compute closed form expressions. This marginal likelihood based methodology, an alternative to currently available methods based on ordinary least squares and method of moments, has monotone estimate of survival function and can accommodate time-varying regression coefficients and covariates. We also present the asymptotic properties of the semiparametric empirical Bayes estimates. We illustrate our methodology using existing statistical software (SAS).
|
