The model is a network comprised of all 3076 counties in the continental US. Each county is a separate node in which epidemics propagate according to an SIR model subject to daily air and auto travel inflows and outflows. Roughly 1 billion long distance auto and 262 million air round trips from county-to-county comprise the travel model. Epidemic parameter values and initial conditions are based on swine flu data from spring 2009. This large scale simulation model is used to simulate US epidemics under different transportation network configurations.
These simulated epidemics behave nothing like the Colizza et al. simulations. Long distance auto travel is the main driver of epidemics contra H3. In contrast to H2, travel essentially disperses infecteds across the country and county infected evolution behaves like an SIR model without travel once the incidence of infection reaches 1/100,000. Per capita travel is not higher in large counties contra part one of H1 but more connected counties experience disproportionately larger travel flows consistent with the second part. However, such hub counties are not preferred epidemic pathways in the simulations.
More effective travel restriction and quarantine policies and targeted pharmaceutical interventions can be designed when large hub airports are the primary epidemic pathways through which epidemics propagate but there are no preferred epidemic pathways if these simulations are correct. The social importance of the resulting inability to materially alter epidemic outcomes in this fashion cannot be overstated.
This project shows that long distance auto, not air, travel is the main epidemic propagation mechanism in the US and that no travel mode is a preferred epidemic pathway that can be used to improve prediction and policy interventions.
Science Fair Project done By Meredith P. Lehmann