PROJECT SUMMARY
Almost half of American adults have a preventable chronic disease, most of which could be improved
with regular physical activity (PA). These proportions are even higher for racially/ethnically diverse populations
where disparities emerge in both chronic disease and PA behavior. Importantly, adults with ~20 minutes/day of
physical activity have a 33% lower risk for all-cause mortality than those who are inactive We know that
physical activity patterns for adults have their developmental beginnings in childhood. Although we know
roughly when, specifically how to affect these patterns is multi-factorial. Direct “cause and effect” models are
insufficient to accommodate the layers of complexity involved in pattern formation. Such complexity includes
multiple dynamic systems with inter- and intra- interactions that influence children’s PA behaviors, including the
built environment, the social environment (both inside and outside the home), and cognitive processes that vary
during- child development. Providing a deeper understanding of these dynamics can advance interventions and
policies for childhood PA behaviors and long-term health disparities reduction.
To accomplish this task, we will leverage approaches more commonly used outside of biomedical
research, in fields such as ecology and social science, and bring together a trans-disciplinary and cross-sector
team of experts in complex systems modeling approaches (Brookings Institution) and pediatric PA and health
disparities (Vanderbilt) to build an etiologic Agent-Based Model (ABM) that identifies which modifiable
determinants may have the biggest impact on PA pattern formation for diverse young children. This project will
utilize an independent dataset collected by the Growing Right Onto Wellness (GROW) Trial of child-parent pairs
to inform the ABM. All of these families represented diverse under-served populations in Tennessee. The GROW
Trial (total N=610 children ages 3-8) collected detailed objective PA data (via accelerometry) at four annual time-
points over the study period (for child-parent pairs), as well as data on the child’s social environment, built
environment, and cognitive processes. Using ABM in this context leverages the diversity and richness of this
longitudinal dataset to build a model with empirically derived parameter estimates to generate new insights into
supporting early childhood PA in diverse populations. ABMs allows us to examine how, when, and for whom
PA behaviors are dynamically shaped by macro-level influences such as the built environment in which
children reside, meso-level influences such as social environments both in and out of the home, and micro-level
influences such as individual cognitive processes in early childhood development. We will examine the potential
heterogeneity in these influences across child characteristics including gender, race/ethnicity, BMI, and BMI
change over time. The result of this project will be a set of data-driven policy recommendations, based
on a complex systems approach to studying childhood PA behaviors, that can be applied in real-world
community settings.