Accurate measurement of free-living physical activity (PA), energy expenditure (EE), and sleep of children (5-12yrs) is
complex, with no single method free of limitations. Validation studies of PA, EE, and sleep have demonstrated that
combining HR and accelerometry data (e.g., steps, counts, raw signal) provides the most accurate estimate of PA, EE,
and sleep. Unfortunately, the simultaneous collection of HR and accelerometry over routine monitoring timeframes (e.g.,
7 days) has been limited because historically measuring HR has relied on uncomfortable chest strap telemetry.
Advancements in wearable technology have eliminated this issue by incorporating the noninvasive assessment of HR
via photoplethysmography in widely-available consumer wearable devices (e.g., FitBits, Garmin) that also include
accelerometry. Studies have shown that HR estimates from consumer wearables are comparable to those collected via
ECG or chest strap telemetry. However, validation studies of consumer wearables have focused almost exclusively on
proprietary activity output (e.g., steps) and have mostly been conducted on healthy adults, older adults, or clinical
populations (e.g., people with neuromuscular or gait abnormalities). Consumer wearables hold promise for collecting
PA, EE, and sleep data with children 5 to 12yrs, yet no studies have been conducted to establish their validity and
utility/feasibility in this population. The objectives of the proposed project are to conduct a series of studies that include
both laboratory and field-based protocols to evaluate the reliability, validity and utility/feasibility of consumer wearables
for measuring children’s PA, EE, and sleep in free-living conditions. We will evaluate the different features of the devices
(e.g., PA, HR) in the lab and in real-world conditions. In addition, we will evaluate the utility/feasibility of consumer
wearables for multi-day wear compliance. We will accomplish the following aims. Aim 1. Develop and validate open-
source equations to estimate PAEE and time spent physically active using the activity and HR data from consumer
wearables compared to the PAEE output from the consumer wearables’ proprietary processing algorithms and a
criterion measure of PAEE (i.e., indirect calorimetry). Aim 2. Develop and validate open-source equations to estimate
total sleep time, sleep efficiency, and timing from consumer wearables using their activity and HR data compared to the
sleep output from the consumer wearables’ proprietary processing algorithms and a criterion measure of sleep (i.e.,
PSG and actigraphy). Aim 3. Evaluate the validity of the PAEE and sleep equation estimates from the algorithms created
in Aim 1 & 2 against a criterion (i.e., Actiheart) under free-living conditions. This project is significant because it will be
among the first to establish the validity of consumer wearables for PAEE and sleep monitoring of children. This project
is innovative as it will use advanced statistical modeling techniques, including machine learning, to systematically test
the validity and utility/feasibility of consumer wearables for children. Our vision is to leverage the biometric data collected
across consumer wearables to produce estimates of PAEE and sleep in children. This will allow practitioners and
researchers alike to more accurately measure 24-hour movement behaviors in children.
