|
Activity Number:
|
133
- Statistical Issues in Environmental Epidemiology and Pharmacoepidemiology
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 9, 2021 : 1:30 PM to 3:20 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
|
Abstract #319166
|
|
|
Title:
|
Bayesian Multiple Index Models for Environmental Mixtures
|
|
Author(s):
|
Glen McGee* and Ander Wilson and Brent Coull and Thomas Webster
|
|
Companies:
|
University of Waterloo and Colorado State University and Harvard TH Chan School of Public Health and Boston University
|
|
Keywords:
|
Environmental health;
Multiple index models;
Kernel machine regression;
Variable selection;
Bayesian methods
|
|
Abstract:
|
An important goal of environmental health research is to assess the risk posed by mixtures of environmental exposures. Two popular classes of models for mixtures analyses are response-surface methods and linear-index methods. Response-surface methods estimate high-dimensional surfaces and are highly flexible but difficult to interpret. Linear-index methods decompose coefficients from a linear model into an overall mixture effect and component weights; these models yield easily interpretable effect estimates and efficient inferences but can be overly restrictive. We propose a Bayesian multiple index model framework that combines the strengths of each, allowing for non-linear and non-additive relationships between exposure indices and a health outcome, while reducing dimensionality and estimating index weights. The proposed framework allows one to select an appropriate analysis from a spectrum of models varying in flexibility and interpretability, and it contains both response-surface and linear-index models as special cases. Unlike fully non-parametric alternatives, the framework also provides a means of incorporating prior knowledge about mixtures in future analyses.
|
Authors who are presenting talks have a * after their name.
