Models on networks

Most models assume that populations are well-mixed, or consider multiple sub-populations in which individuals within each sub-population are assumed to mix randomly with other members of the same sub-population. While increasing the number of sub-populations can accommodate heterogeneity in contacts to a certain degree, it may be more biologically realistic to assume that individuals contacts are structured as a network, which may be static or may change over time.

One approach to modeling the spread of infection over a network is to use detailed simulations. This approach is used in the R package EpiModel. Alternatively, compartmental models that approximate the dynamics on a network can be used, such as pair approximations or edge-based models.