Activity Number:
|
529
- Contributed Poster Presentations: WNAR
|
Type:
|
Contributed
|
Date/Time:
|
Wednesday, July 31, 2019 : 10:30 AM to 12:20 PM
|
Sponsor:
|
WNAR
|
Abstract #304831
|
|
Title:
|
An Adjusted Partial Least Squares Regression Framework for Environmental Mixture Data Analysis
|
Author(s):
|
Ruofei Du* and Timothy Ozechowski
|
Companies:
|
University of New Mexico Comprehensive Cancer Center and University of New Mexico Health Sciences Center
|
Keywords:
|
environmental mixture data;
low signal to noise ratio;
partial least squares regression ;
adjusted partial least squares regression
|
Abstract:
|
There are many challenges in statistical analysis of environmental mixture data. One of them is the low signal to noise ratio in the association between the mixture and an adverse outcome. It’s often seen some association detected in an unreasonable direction. A statistical method needs to be sensitive in capturing the weak signal, and capable of using all available information. The partial least squares regression (PLSR) creates the components by maximizing the correlation between the mixture and outcome simultaneously. The data is rotated in favor of detection the association. In real studies it’s also often observed that only small percentage of participants have outcome measurements. By a general PLSR approach, the information of the participants who don’t have outcomes are not able to be utilized in the analysis. We propose the methodology adjustments on the general PLSR approach so that the information of all the participants will contribute on the model fitting and the assessment of the association effect. In the simulation study, the proposed framework leads to a superior performance in both estimation and hypothesis testing of the association effect.
|
Authors who are presenting talks have a * after their name.