Bootstrapping in the Cox-model with interval censored observations

Master thesis (2019)

Authors

S. Gouwens Electrical Engineering, Mathematics and Computer Science

Contributors

G. Jongbloed Delft Institute of Applied Mathematics (mentor)
Faculty
Electrical Engineering, Mathematics and Computer Science, Electrical Engineering, Mathematics and Computer Science
Copyright: © 2019 Sigur Gouwens
More Info
expand_more

Published Date

01-12-2019

Language

English

Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Faculty

Electrical Engineering, Mathematics and Computer Science

Abstract

In this study the interval censoring case 2 model combined with the Cox model is considered. The event-time distribution function is modeled nonparametrically. Two algorithms are proposed to estimate the event-time distribution function together with the Cox coefficients. Kernel smoothing is applied to the non-parametric MLE of the event-time distribution resulting in the smoothed MLE (SMLE). A two-step method for choosing the smoothing bandwidth based on minimising the MSE is introduced. Given the SMLE, the precision of the MLE is tested using bootstrap simulations. New event-times are sampled from the SMLE which are then used to compute bootstrap estimates of the event-time distribution. This is done for multiple sample sizes to observe large sample behaviour. This study suggests that larger sample sizes lead to
better estimates. Monte Carlo simulations and the bootstrap simulations agree on the bandwidth and the large sample distribution of pointwise estimates of the event-time distribution.