|
Activity Number:
|
474
- SPEED: Infectious Disease, Environmental Epidemiology, and Diet
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2018 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
|
Abstract #330079
|
|
|
Title:
|
Creating a Composite Score for Physical Activity Using Shape Constrained Additive Model
|
|
Author(s):
|
Eli Kravitz* and Raymond J. Carroll and Sarah Keadle
|
|
Companies:
|
Texas A&M Statistics and Texas A & M University and California Polytechnic State University
|
|
Keywords:
|
epidemiology;
nonparametric regression;
generalized additive models;
physical activity;
composite score
|
|
Abstract:
|
In public health fields, composite scores are commonly used to assess a health behavior. These composite scores compare an individual's health behavior to an idealized standard and provide a number between 0 and 100 to indicate their health status. We add to the literature on this topic by providing a composite score for physical activity. Using the NIH-AARP Study of Diet and Health, we break physical activity into 8 discrete components. We use Generalized Additive Models to describe the relationship between these 8 components and mortality. We have additional information that can be incorporated into our model: each of the smooth functions must satisfy certain shape constraints. For example, kinesiologists have shown that vigorous activity is beneficial to overall health but the benefit asymptotes. Translated mathematically, this relationship is be concave and increasing. We enforce similar constraints on each of the smooth functions. We fit this shape constrained GAM and apply simple transformations to the fitted values to create a number between 0 and 100 that is still predictive of mortality. We provide a simulation procedure for testing the validity of our results.
|
Authors who are presenting talks have a * after their name.
