A network-driven approach to modeling the spread of Ebola-type epidemics

We propose new models for the spread of Ebola-type epidemics, considering networks both within and between countries. By modifying the traditional SIR model to capture the effects within each specific country, our Spatial SI(D/S) model overlays geographic information in the form of a graph topology to model the spread of diseases across boarders. In fitting the models to real-world data from the 2014–16 West Africa Ebola outbreak, we find that each is able to obtain low error in predicting infections over time, and that the use of spatial information can provide at least marginal improvements. We also show how our model parameters offer more insights into how these types of diseases are spread than does the SIR model, and propose an optimization problem for epidemic response strategies on a fixed budget that makes use of these parameters.