|
Abstract:
|
Analysis of health effects of exposure to environmental chemical mixtures poses various problems to researchers. These problems are often related to dimensionality of the potential exposures of interest, the complex correlation structure in the exposures, high uncertainty in the measurements of the exposures, possible non-linear and interacting relationship between the exposures and the health endpoints, the presence of confounders, and difficulty of interpreting the results of statistical models. In an attempt to resolve these issues, we propose a two-stage approach that can be applied to the analysis of health effects associated with environmental chemicals. In the first stage of our approach, we propose to reduce the dimensionality of the exposure variables using a novel informed sparse principal components analysis. In the second stage of the approach, we propose to analyze the effects of these lower dimensional exposure variables using a segmented linear regression analysis.
|
