Accurate assessment of children’s (5-12yrs) free-living physical activity energy expenditure (PAEE) is critical to
understanding the complex and interdependent relationship between children’s PAEE and health outcomes. The
combination of movement (e.g., steps, counts, raw signal) and heart rate data provides reasonably accurate
estimates of children’s PAEE during lab conditions. However, the combination of movement and heart rate
(MOVE+HR) may be inadequate for predicting children’s free-living PAEE. Studies in adults demonstrate
including the type of activity (e.g., the participant is walking, running, cycling, etc.) improves estimates of free-
living PAEE, relative to using movement or MOVE+HR. Yet, this evidence is based on studies conducted in
adults. No studies of children have investigated the impact of adding activity type to PAEE prediction equations
that use MOVE+HR. Existing consumer wearables (e.g., Fitbit, Garmin) are promising for the assessment of
children’s free-living PAEE. These devices incorporate built-in pattern-recognition features that automatically
detect activity type, and these metrics are provided in a user-friendly format for the end-user; however, no known
studies have evaluated the ability of consumer wearables to autodetect activities in children. Consumer
wearables also use photoplethysmography to capture HR and accelerometry to capture movement. Their unique
ability to capture activity type, HR, and movement metrics could significantly improve estimates of free-living
PAEE in children. Aim 1 will evaluate the impact of including activity type, captured via direct observation, to
regression equations that use movement and HR data from consumer wearables (i.e., Garmin Vivoactive 4S
and Fitbit Sense). Aim 2 will evaluate the ability of consumer wearables (i.e., Garmin Vivoactive 4S and Fitbit
Sense) to automatically detect activities (i.e., walking, running, biking). To accomplish these aims, this study will
capitalize on the validation design (i.e., semi-structured physical activity protocol and data
management/extraction procedures) and draw on data collected from 120 children (5-12yrs) from an existing
R01 project. This project will use analytical techniques, including cross-sectional time series (CSTS), multivariate
adaptive regression spline (MARS), machine learning, and equivalence testing to address the following aims.
The project’s long-term goal is to advance the assessment of children’s PAEE in epidemiologic- and intervention-
based studies for children, which is critical to understanding the complex relationship between children’s PAEE
and health outcomes. Through the execution of this project, the following will be gained: an in-depth knowledge
of the literature related to assessment of PAEE in children, expertise designing and implementing validation
studies assessing PAEE in children, hands-on training using gold standard measures to assess children’s PAEE
and activity type, proficiency performing advanced analytical techniques, and scientific communication and
grantsmanship skills, including peer-reviewed publication, scientific presentation, mentored manuscript review,
and a drafted post-doctoral grant application.