Abstract: Humans are routinely exposed to mixtures of chemical and other environmental factors, making the quantification of health effects associated with environmental mixtures a critical goal for establishing environmental policy sufficiently protective of human health. The environmental health community has recognized the development of statistical methods to deal with unique challenges associated with environmental mixtures data as a priority in the field. A growing body of literature has developed and applied a wide range of statistical tools for characterizing a health effects of an environmental mixture, including methods for variable selection, dimension reduction, and statistical learning more broadly, to address important environmental health problems. The majority of this literature has focused on approaches for data collected on a set of exposures measured at a single point in time, for a single outcome. However, modern environmental epidemiology seeks to assess risk in settings much more complex than this relatively simple scenario. Examples include assessment of critical windows of exposure in children, health effects on outcome trajectories, and mediation analyses, to name just a few. In this talk I will summarize some of our recent efforts to fill this gap, describing Bayesian machine learning approaches to quantify the joint impacts of metal mixture exposures on birth outcomes and neuro-development in children from Bangladesh and Mexico.
Integrated Health Sciences Core of the Michigan Center on Lifestage Environmental Exposures and Disease (M-LEEaD)Statistical Methods for Modern Mixtures Epidemiology: Windows of Susceptibility, Longitudinal Trajectories, and Mediation - Brent Coull
A special event of the Integrated Health Sciences Core
March 14, 2018
11:30 a.m. - 1:00 p.m.
1655 SPH I
1415 Washington Heights
Ann Arbor, MI 48109-2029
Sponsored by: Integrated Health Sciences Core of the Michigan Center on Lifestage Environmental Exposures and Disease (M-LEEaD)
Contact Information: Meredith McGehee (mcgehee@umich.edu | 647-0819)
Abstract: Humans are routinely exposed to mixtures of chemical and other environmental factors, making the quantification of health effects associated with environmental mixtures a critical goal for establishing environmental policy sufficiently protective of human health. The environmental health community has recognized the development of statistical methods to deal with unique challenges associated with environmental mixtures data as a priority in the field. A growing body of literature has developed and applied a wide range of statistical tools for characterizing a health effects of an environmental mixture, including methods for variable selection, dimension reduction, and statistical learning more broadly, to address important environmental health problems. The majority of this literature has focused on approaches for data collected on a set of exposures measured at a single point in time, for a single outcome. However, modern environmental epidemiology seeks to assess risk in settings much more complex than this relatively simple scenario. Examples include assessment of critical windows of exposure in children, health effects on outcome trajectories, and mediation analyses, to name just a few. In this talk I will summarize some of our recent efforts to fill this gap, describing Bayesian machine learning approaches to quantify the joint impacts of metal mixture exposures on birth outcomes and neuro-development in children from Bangladesh and Mexico.
