Disease
spread in most biological populations requires the proximity of agents.
In populations where the individuals have spatial mobility, the contact
graph is generated by the “collision dynamics” of the agents, and thus
the evolution of epidemics couples directly to the spatial dynamics of
the population. We first briefly review the properties and the
methodology of an agent-based simulation (EPISIMS) to model disease
spread in realistic urban dynamic contact networks. Using the data
generated by this simulation, we introduce the notion of dynamic
proximity networks which takes into account the relevant time-scales for
disease spread: contact duration, infectivity period, and rate of
contact creation. This approach promises to be a good candidate for a
unified treatment of epidemic types that are driven by agent collision
dynamics. In particular, using a simple model, we show that it can
account for the observed qualitative differences between the degree
distributions of contact graphs of diseases with short infectivity
period (such as air-transmitted diseases) or long infectivity periods
(such as HIV).
