|
Activity Number:
|
369
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 6, 2008 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #301642 |
|
Title:
|
Using Semiparametric Varying Coefficient Models To Investigate Interactions of Toxic Exposure and Nutritional Covariates in a Study of Neurodevelopmental Outcomes
|
|
Author(s):
|
Miranda L. Lynch*+ and Li-Shan Huang
|
|
Companies:
|
University of Rochester and University of Rochester
|
|
Address:
|
Department of Biostatistics and Computational Biology, Rochester, NY, 14642,
|
|
Keywords:
|
Varying coefficient models ; Semiparametric models ; Neurodevelopmental outcomes
|
|
Abstract:
|
Varying coefficient models are a flexible method for quantifying interactions by allowing the linear association between a primary regressor and a response to be modeled via linear model coefficients that are themselves smooth functions of a separate effect-modifying covariate. The Seychelles Child Development Study (SCDS) is a long-term study of the effects of prenatal mercury (Hg) exposure on measures of neurodevelopmental endpoints in children. Recent work in the SCDS has suggested that deleterious effects of Hg on development may be attenuated by complicated interactions of Hg with one or more nutrition covariates. Interpretation of these interactions is crucial to understanding the role of nutrition in modifying the effects of Hg on developmental outcomes. These interactions will be explored using semiparametric varying coefficient models based on penalized splines.
|
