|
Activity Number:
|
116
- Epidemiological Models for Longitudinal Studies, Time-to-Event Outcomes, and Functional Data
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 8, 2022 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
|
Abstract #322865
|
|
|
Title:
|
Functional Regression Calibration and Its Application to NHANES 2003-2006
|
|
Author(s):
|
Lucia Tabacu *
|
|
Companies:
|
Old Dominion University
|
|
Keywords:
|
NHANES 2003-2006;
accelerometry ;
physical activity ;
measurement error ;
mortality prediction ;
functional data
|
|
Abstract:
|
We study the scalar-on-function regression model where the latent functional covariate X(t) is not directly observable but measured by a proxy W(t) with measurement error U(t), possibly correlated over time with unspecified structure. Thus, we assume an additive measurement error model with replicates for the proxy variable W(t). Through simulation studies we compare the parameter estimation when the functional covariate X(t) is replaced by the average of the replicates W(t) and when X(t) is replaced by the linear approximation to X(t) given other covariates and the replicate measurements (Carroll and Stefanski 1990; Gleser 1990). As an application we consider the mortality prediction in the National Health and Nutritional Examination Survey (NHANES) 2003-2006, where the latent functional covariate is physical activity measured every minute for 7 days by a hip-worn accelerometer. This is joint work with Andrew Leroux (CU) and Ciprian Crainiceanu (JHU).
|
Authors who are presenting talks have a * after their name.
Back to the full JSM 2022 program
|
