Activity Number:
|
388
|
Type:
|
Contributed
|
Date/Time:
|
Thursday, August 15, 2002 : 8:30 AM to 10:20 AM
|
Sponsor:
|
Biometrics Section*
|
Abstract - #300843 |
Title:
|
On Estimating the Gamma-Accelerated Failure-Time Models
|
Author(s):
|
Kallappa Koti*+
|
Affiliation(s):
|
Food and Drug Administration
|
Address:
|
1401 Rockville Pike, Suite 200S, HFM-219, Rockville, Maryland, 20852, USA
|
Keywords:
|
Incomplete gamma function ; Fisher information matrix
|
Abstract:
|
We discuss the computational issues involved in estimating the gamma-accelerated failure-time model using maximum likelihood. In particular, we propose a trapezoidal rule-based hybrid approximation for the derivative of the log likelihood, with respect to the shape parameter. We use the SAS/IML subroutine NLPTR to obtain the maximum likelihood estimates of the model parameters. We also compute the variance-covariance matrix of the parameter estimators. We illustrate the procedure using the leukemia data from Kersey, et al. (1987) and the pediatric cancer data from Cantor and Shuster (1992). The leukemia data are 24 percent censored, whereas the pediatric cancer data are 87 percent censored. We compare our results with those obtained using the SAS LIFEREG procedure. We observe that the SAS LIFEREG procedure performs poorly when applied to the Cantor-Shuster data, and it overestimates the standard error of the shape parameter for the Kersey's data. We are inclined to claim that our approach gives better results in both cases.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2002 program |