(A pirate copy of the paper is here.)
The paper describes strategies for reducing the spread of epidemics of infectious diseases that are transmitted by close contact between people, even when transmission can occur before symptoms appear, as in flu. The basic idea is social distancing: reducing close contact between people. The strategies are designed to decide which contacts should be avoided, to have the greatest effect on the epidemic relative to the number of contacts avoided.
All strategies exploit the fact that the contact network has community structure: groups of people have more close contact with each other than with members of other groups. Infection spreads faster within communities than between them, which explains why emerging diseases affect certain regions long before spreading elsewhere. The same applies to communities at smaller scales (towns, buildings, offices, schools, households, etc.). Our strategies are local, meaning that individuals decide which contacts to avoid, based on their local views of the contact network. There is no need for a totalitarian authority to decide which contacts should be avoided (e.g., by quarantining people).
The paper describes several scenarios. The one most relevant to widely known diseases is the global belief-based model (Section III.A), in which the whole population becomes aware of the disease at an early stage and applies social distancing according to various strategies. The other models, which involve waiting for symptoms to appear or for awareness to spread, may be too slow to be effective.
The paper tested several social distancing strategies, which decide which contacts to avoid. The most effective overall was the strategy based on local modularity. To allow this strategy to be used, we have implemented a social distancing tool (below). The tool allows anyone to specify their own contact network, to find out which contacts should be avoided. The experiments in the paper assumed that everyone in the population uses the same strategy repeatedly to find several (a fixed fraction) of their existing contacts to cut contact with.
| The tool also includes an example network, which is interesting because the strategy recommends avoiding a contact that is not the one with the greatest number of contacts. |
Steve Gregory
Email: stevegregory2@gmail.com