Nonparametric maximum likelihood estimation for semicompeting risks data with validation sampling.

• Main Authors: LIN, CHIEN-HSU, 林建旭
• Other Authors: HUANG, CHIA-HUI
• Format: Others
• Language: zh-TW
• Published: 2015
• Online Access: http://ndltd.ncl.edu.tw/handle/95878162481933420139

碩士 === 國立臺北大學 === 統計學系 === 103 === The semicompeting risks model with validation sampling is considered to study the covariates effect through the illness-death model, where crude covariate information is available from the individuals while true covariate information is collected from part of sample. The one-dimensional Cox model under the validation sampling has been studied by Chen (2002). However, the idea and methodology can be applied to the multiple events such as semicompeting risks data. In this work, it is assumed that the occurrence of nonterminal event is known at the observed terminal event time and undefined when terminal time is censored. It is of interest to make use both crude and true covariate information to improve the efficiency of the estimates. An extension of Chen’s method has been proposed to the data of this type. The extensive simulation studies are established to study the large sample properties of the maximum likelihood estimates and their results are provided to illustrate the performance of the proposed method.