|
Activity Number:
|
322
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 2, 2016 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
|
Abstract #321152
|
View Presentation
|
|
Title:
|
A Semiparametric Bayesian Survival Analysis Using Covariate-Dependent Clustering, with Application to Pediatric Kidney Transplantation Data
|
|
Author(s):
|
Hang J. Kim* and Mi-Ok Kim
|
|
Companies:
|
University of Cincinnati and Cincinnati Children's Hospital Medical Center
|
|
Keywords:
|
risk factor ;
probit model ;
Bayesian analysis ;
causal analysis ;
clustering ;
survival model
|
|
Abstract:
|
Clustering approaches can be useful in exploring for incompletely understood risk factors and treatment interaction effects in biomedical studies. In this abstract we illustrate a Bayesian risk-factor-dependent clustering approach. A motivating example comes from a pediatric kidney transplant study where a pre-identified set of risk factors exists that potentially interact with the effect of steroid-avoidance protocols (treatment) on the graft survival (response), whereas non-randomized treatment selection and potentials of high order interactions complicates identification of predictive factors as opposed to prognostic factors, let alone exact forms of the treatment interactions. We used propensity score matching to account for confounding bias and risk-factor-dependent clustering of the subjects to identify predictive factors. By applying the suggested approach to the pediatric kidney transplantation data, we explain the different impact of treatments on survival by subgroups as well as specify the characteristics of the particular subgroups for which steroid-avoidance protocols are effective.
|
Authors who are presenting talks have a * after their name.
