Nonparametric Estimation of Conditional Quantile Residual Life Under Censoring

• Activity Number: 627
• Type: Contributed
• Date/Time: Thursday, August 2, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Wen-Chi Wu*+ and Jong-Hyeon Jeong
• Companies: University of Pittsburgh and University of Pittsburgh
• Keywords: residual life ; bivariate survival function ; kernel density estimator ; Kaplan-Meier estimator

Abstract

Inference on the remaining lifetime of a patient is of interest in many clinical studies. Numerous studies for estimating the quantile residual life function have been conducted in the univariate settings with or without covariates. However, some patients may experience two different types of events and the conditional quantile residual life time to a later event after experiencing the first event might be of utmost interest, which requires a bivariate modeling of quantile residual lifetimes subject to right censoring. In the bivariate survival data, observed survival times could be singly or doubly censored, or both uncensored. Under a bivariate setting, the conditional survival function given one of random variables can be estimated by the Kaplan-Meier estimator with a kernel density estimator and bandwidth which provides different weights for censored and uncensored data points around the region in a plane (M. J. van der Laan, 1997). A quantile residual lifetime can be then estimated at a fixed time point by inverting the estimated conditional survival function. The asymptotic properties of the conditional quantile residual life function estimator will be investigated both theor