Penalized Nonparametric Likelihood-based Inference for Current Status Data Model

Abstract

In this paper, we develop a penalized nonparametric likelihood method to estimate an unknown cumulative hazard function with current status data. Deriving the limiting distribution of such nonparametric estimator is a very challenging theoretical problem. For the problem, we construct the Sobolev space equipped with a special inner product and deduce a functional Bahadur representation in the space. Using this key tool, we establish the pointwise asymptotic normality of the proposed estimator.

Furthermore, we study the penalized likelihood ratio tests for local and global ypotheses and obtain their limiting distributions, and also show the optimality of the test. A simulation study is presented for comparing the performance of the proposed penalized likelihood ratio test and the classical likelihood ratio test.

Biography

Prof. Xingqiu Zhao received her PhD at McMaster University and is currently an associate professor at Hong Kong Polytech. Her main research interests are panel count data, Longitudinal data analysis and large deviation with applications in survival analysis. She has published more than 50 papers on international journals such as Annals of Statistics, JASA, Bernoulli and so on.