Project Summary
Coronavirus Disease 2019 (COVID-19) continues to cause significant morbidity and mortality across the world.
Characterized by crowded detention facilities and limited medical safety resources, US criminal legal settings
(CLS) have experienced some of the largest COVID-19 outbreaks. Persons involved with CLS (PCLS)
additionally experience significant barriers to health care upon release, and often return to environments
impacted by syndemic factors rooted in structural racism: lower vaccine access, fewer testing facilities, medical
mistrust, and higher COVID-19 prevalence. A complex interplay between individual and social network-level
factors may be driving the adverse COVID-19 outcomes among PCLS. But, despite their importance, social
network influences – and the effect of their interaction with individual ands structural factors on COVID-19
testing, vaccination and broader health behaviors – are not routinely examined. To systematically address this
gap, we will leverage two existing RADx-UP studies across eight US states. The “Community Network Driven
COVID-19 Testing Among Most Vulnerable Populations in the Central United States” (C3) study is unique in
that it has collected longitudinal social network data on testing, vaccination and health behaviors among PCLS
in five US states. Additionally, the “COVID-19 Testing and Prevention in Correctional Settings” (CTC) study has
assessed COVID-19 testing, vaccination, and mitigation strategies for PCLS in three US states. We will
integrate the common data elements collected through the CTC project with the network determinants
estimated from the C3 data to develop an agent-based network model (ABNM) – a dynamic systems modeling
technique that provides the ability to simulate emergent interaction between individual behaviors, social
structures, policy implementation, and downstream assessment of population outcomes. The proposed
modeling study will: (1) use machine learning to quantify the impact of network-level influences on COVID-19
testing, vaccination, and health behaviors within PCLS communities; (2) build an agent-based network
modeling (ABNM) platform that integrates the individual common data elements (CDEs) of testing and
vaccination collected from the CTC study and network determinants from the C3 study; (3) simulate the effects
of interventions on COVID-19 vaccination, testing and broader health behaviors in PCLS and their
communities. This approach will provide insight on the potential impacts of network-informed interventions
using RADx-UP data, social network analysis, machine learning, and agent-based modeling to identify
interventions to reduce COVID-19 morbidity and mortality among PCLS and their communities.