|
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 #318300
|
|
|
Title:
|
Partial-Linear, Single-Index Quantile Regression for Modeling Time-Dependent Environmental Exposures
|
|
Author(s):
|
Yuyan Wang* and Mengling Liu
|
|
Companies:
|
NYU Langone Health and New York University Grossman School of Medicine
|
|
Keywords:
|
Time-dependent exposures;
multiple exposures;
distributed lag;
functional effects;
tail quantiles
|
|
Abstract:
|
Studying time-dependent environmental exposures is of great importance to evaluate their time-varying impacts on human health. When health outcome has skewed distribution or its tail quantiles are of interest, quantile regression is widely applied. We thus propose partial-linear single-index distributed-lag (PLSI-DL) quantile regression (QR) models to evaluate the effects of time-dependent environmental exposures on the different quantiles across the distribution of a time-invariant health outcome. We consider two settings on how time-dependent exposures are measured, including finite measurements within fixed windows and dense measurements as functional exposures. Spline methods are used to estimate the effect function for time-dependent exposures and the single index link function, respectively. We will demonstrate the performance of our proposed method through numerical studies in diverse scenarios and applications in birthweight data for discrete and functional environmental pollutant exposures.
|
Authors who are presenting talks have a * after their name.
