Nonparametric Regression for Correlated Failure Time Data

Zhang-sheng Yu
Biostatistics
University of Michigan
*Candidate for the Assistant or Associate Professor Position

Wednesday, February 22nd
3:30pm
Moos 2-620
Minneapolis Campus

Abstract:
We first study nonparametric regression for the correlated failure time
data under the marginal proportional hazard model. Kernel estimating
equations are used to estimate nonparametric covariate effects.
Independent and weighted working kernel estimating equations (EE) are
studied. We show that the nonparametric estimator of the covariate
function's derivative is consistent for any arbitrary working correlation
matrix and the asymptotic variance is minimized by assuming working
independence. We evaluate the performance of the proposed kernel estimator
using simulation studies.

We also propose a double penalized partial likelihood (DPPL) for Gaussian
frailty model with nonparametric regression using smoothing spline. This
provides a unified approach for nonparametric covariate function and
smoothing parameter estimation. The variance component estimation method
(ML) were used to estimate the smoothing parameter. Numerical analysis
shows that ML estimation works well for smoothing parameter estimation.
The point estimator and standard error estimator of covariate function
performs well. We apply the proposed method to western Kenya parasitemia
study.
--------------------------------------------
A social tea will be held at 3:00P.M. in A434 Mayo. All are Welcome.
For more details contact 612-624-4655 or see
http://www.biostat.umn.edu/seminar_academic.html