SEMINARS
|
|
Fall 2007
STATISTICS
COLLOQUIUM
Wednesday, November 12, 2008
3:30-4:00—Refreshments
4:00-5:00—Talk
Yost Hall, Room 101
Bo Hu, Ph.D.
Assistant Staff
Cleveland Clinic Foundation
A Semiparametric Joint Model for Longitudinal and Survival Data
In many clinical studies, the repeated measurements of some biomarkers on each subject quantify the severity of the disease and that subject's susceptibility to progression of the disease.
It is of scientific interest to relate such quantities to a later clinical endpoint such as patient survival. This type of data is usually analyzed by the joint modeling approach, which assumes that the distributions of the longitudinal data and the survival data are independent, conditioning on some shared parameters (frailties). We consider a general form of the shared parameter model. A corrected joint likelihood approach was proposed for estimation of parameters. The corrected log likelihood is shown to be asymptotically concave and leads to consistent and asymptotically normal estimators. The estimation procedure does not rely on distributional assumptions of the frailties. We also allow time-dependent covariates in the model. The proposed method was studied in simulations and applied to a data set from the Hemodialysis (HEMO) Study.
|
