Proportional hazard model estimation under dependent censoring using copulas and penalized likelihood – an application to a dementia dataset

Thursday, 28 June 2018
Melbourne School of Population and Global Health
207 Bouverie St, Carlton, Level 5, Room 515

This talk starts with the problem of Cox’s proportional hazards model estimation for a dementia dataset where informative right censoring is likely to exist. We adopt the method of maximum penalized likelihood, where dependence between censoring and event time is modelled by a copula function and a roughness penalty function is used to restrain the baseline hazard as a smooth function. We will discuss how to fit this model using a special constrained optimization method and then will also explain the asymptotic properties for both regression coefficients and baseline hazard estimates. The simulation study is conducted to investigate the performance of our method and also compares it with an existing maximum likelihood method. Finally, we apply the proposed method to the dementia patients dataset discussed at the beginning.

A/Prof Jun Ma

Assoc. Prof. Jun Ma

Department of Statistics
Macquarie University

A/Prof Jun Ma works on semiparametric regression models for survival analysis, particularly Cox models, additive hazard models and accelerated failure time models where survival data include event times as well as left, right or interval censoring times. Jun and his collaborators and PhD students have developed efficient constrained optimization algorithms to estimate the parameters of these models, including estimation of the “nonparametric” components. They have also solved the difficult problem of asymptotic covariance matrix computation where active constraints are taken into consideration. Jun has also worked on other research projects, such as imaging processing/reconstruction, lasso model selection, empirical likelihood methods and measurement errors in regression models. 

PDF icon Slides_Jun Ma.pdf3.06 MB