Activity Number:
|
358
- Contributed Poster Presentations: Biometrics Section
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, July 31, 2018 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Biometrics Section
|
Abstract #328392
|
|
Title:
|
Firth Adjustment for Parametric Current-Status Survival Analysis
|
Author(s):
|
Hung-Mo Lin* and JOHN M WILLIAMSON and HAE-YOUNG KIM
|
Companies:
|
Icahn School of Medicine at Mount Sinai and Centers for Disease Control and Prevention and New York Medical College
|
Keywords:
|
bias;
current status data;
parametric survival analysis
|
Abstract:
|
Analysis of right-censored data is often problematic due to infinite maximum likelihood estimates and biased estimates when they do exist, especially for small sample sizes. The analysis of current-status survival data is especially troublesome because of the extreme loss of precision due to the large failure intervals. We extend Firth's method (1993) for reducing bias in regular parametric problems to parametric current-status survival modeling. Firth advocated a general method for reducing the bias of maximum likelihood parameter estimates by systematically correcting the score equation rather than the maximum likelihood estimate itself. An advantage of Firth's approach is that it is still applicable when the maximum likelihood estimate does not exist. We present simulation studies with exponential- and Weibull-distributed data to detail the performance of our approach, as well as 2 illustrative analyses: one of lung tumor data from RFC mice and the other a simulated data set involving separation.
|
Authors who are presenting talks have a * after their name.