contact model decay function

[mikkokotila]

In terms of the contact model, we should probably consider that as time goes by, the chances that the contact is the same, will increase (even in a bus, you tend to go same time to work, and you tend to have more redundant contacts).

[jampekka]

This is a common (and well acknowledged) issue in almost all epidemic models (known as homogenous mixing assumption), and there don't seem to be simple ways of doing it with sufficient computational efficiency. For dynamical systems it can be somewhat approximated using an exponent in the contact frequency.

But I think in an agent-based model this could be relatively simple as we model the progression over time, and we could even give some estimates for it using eg. the POLYMOD's contact frequency data. I think this is worth giving a go as a potential alternative to modeling a full clustered network.

[mikkokotila]

As @jampekka had mentioned, it gets complicated fast. I suggest we don't try to address the decay function at this point. It seems that locality is closely related to this, which at some point would have to be its own model.

But I think in an agent-based model this could be relatively simple as we model the progression over time, and we could even give some estimates for it using eg. the POLYMOD's contact frequency data. I think this is worth giving a go as a potential alternative to modeling a full clustered network.

+1