Robust and Semiparametric Statistical Modeling for Cancer Research.

Robust statistics;Semiparametric methods;Binary regression;Survival analysis;Semicontinuous data;Public Health;Statistics and Numeric Data;Health Sciences;Science;Biostatistics

Rice, John D.Nan, Bin ;

University of Michigan

关键词: Robust statistics;    Semiparametric methods;    Binary regression;    Survival analysis;    Semicontinuous data;    Public Health;    Statistics and Numeric Data;    Health Sciences;    Science;    Biostatistics;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/113285/jdrice_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

In the application of biostatistical methodology to cancer studies, there is a desire to use methods with fewer or less restrictive assumptions, which often lead to more easily generalizable conclusions.The first chapter deals with robust modeling of binary responses with the goal of improving classification at an arbitrary probability threshold dictated by the particular application. Specifically, for thelinear logistic model, we solve a set of locally weighted score equations, using a kernel-like weight function centered at the threshold. This work has much in common with robust estimation, but differs from previous approaches in this area in its focus on prediction, specifically classification into high- and low-risk groups. Analysis of a melanoma data set is presented to illustrate the use of the method in practice.The second chapter addresses the difficulties inherent in investigating time to cancer onset when only time to diagnosis can be observed. To address this problem, we propose a joint model for the unobserved time to the latent and terminal events, with the two events linked by the baseline hazard. We propose an EM algorithm for estimation of the baseline hazard, which allows for closed-form Breslow-type estimators at each iteration, reducing computational time compared with maximizing the marginal likelihood directly. We demonstrate use of the method with analysis of a prostate cancer data set from SEER.In the third chapter, we apply methodology originally used in survival analysis to model semicontinuous data. Continuous outcome data with a proportion of observations equal to zero arises frequently in biomedical studies. We propose a semiparametric model based on a biological system with competing damage manifestation and resistance processes. This allows us to derive a partial likelihood based on the retro-hazard function, leading to a flexible procedure for modeling continuous data with a point mass at zero. We apply the method to a data set consisting of pulmonary capillary hemorrhage area in lab rats subjected to diagnostic ultrasound.