Statistical Methods in Disease Risk Assessment and Prediction

点击翻译按钮获取中文摘要

There are two major accomplishment this year. First, we developed a nonparametric and a semiparametric regression approach for bivariate failure time outcomes. This is motivated by the fact that many biomedical studies follow participants for multiple correlated health outcomes. Modeling these outcomes simultaneously can be more efficient than individual models, and allows us to characterize the risk of developing multiple diseases to facilitate risk prediction given individuals history of other diseases. In addition to two journal manuscripts describing these approaches, we also published a book by Chapman & Hall on this topic. Second, we developed an approach to handle assay limit of detection in environental mixture studies. Conventional approaches to deal with coveriates subject to LOD, including complete-case analysis, substitution methods and parametric modeling of the covariate distribution, are feasible but may result in efficiency loss or bias. We considered a multivariate accelerated failure time model for the multiple correlated covariates subject to LOD and a generalized linear model for the outcome. A two-stage procedure based on semiparametric pseudo-likelihood is proposed for estimating the effects of environmental mixtures on the health outcome. We illustrate the practical utility of the methodology with the LIFECODES birth cohort data, where we compare our approach to existing approaches in analysis of multiple urinary trace metals in association with oxidative stress in pregnant women.