PROJECT SUMMARY
Building realistic models for disease transmission is deeply challenging. While classic epidemiology models
are powerful for building general insight, real epidemics rarely meet model assumptions. Network-based
models promise greater accuracy, but depend on data that are difficult and expensive to collect and simula-
tion toolkits that either make the same simplifying assumptions of classic models or build realism in largely
ad hoc ways. Here we leverage community science to build a collective repository for network data and sim-
ulation code to allow multiple investigators easier access to a wider and more diverse data foundation. In
Aim 1, we will build a secure-but-shareable repository of village and rural community networks to serve as a
general (and ultimately expandable) resource for disease simulation modeling. Village network data have
been collected by multiple investigator teams, with an initial literature search yielding over 1000 village net-
works covering over 80,000 unique people. We will regularize the data structure and harmonize data con-
tent to provide a composite contact-propensity score necessary for modeling disease. Rural settings are at
significant risk of zoonotic spillover and are thus a key study setting to understand early-phase outbreaks
that could lead to future pandemics. Our repository leverages advances in secure data storage and sharing
that allow us to automate the production of data use agreements for sharing and distributing sensitive data.
In Aim 2, we similarly extend the model toolkit by (a) providing tools to automatically link deposited data to
extant disease simulation packages (EpiModel) and (b) providing an extendable network agent-based simu-
lation codebase that can model multiple types of pathogens, agent responsiveness to disease, competing
information flow, and policy interventions. The toolkit will be modular, with multiple examples so that it is
easy to use and customizable for investigator-driven analyses of disease spread for multiple types of patho-
gens, accommodating dynamic changes to the network as a result of agents’ real-time behavioral adapta-
tions to disease and information flow. In Aim 3, we will run two illustrative computational experiments that
create the foundation for a future substantive R01 on the conditional effects of behavior responsiveness by
network position and overall structure. First, we will examine how different levels of disease-relevant cau-
tious behavior - avoiding contact to stay safe - interact with network cohesion to shape disease trajectories.
Second, we will examine how multiple information and disinformation flows compete over real networks and
how their effect on behavior promotes disease spread. This project will make significant advances in the re-
search community’s ability to build realistic models that integrate actor attributes, interests and behavior
with disease and network structure for understanding and predicting future outbreaks by providing a public
resource that can leverage past data collection and modeling investments in new ways.