search

tapios / risk-networks

Code for risk networks: a blend of compartmental models, graphs, data assimilation and semi-supervised learning
Other
2 stars 2 forks source link

Exogenous infections #84

Open odunbar opened 4 years ago

odunbar commented 4 years ago

In a recent meeting it was discussed to include Exogenous infections, where users of the app can be infected by those without it. This could be treated by

  1. Obtaining the number of external connections subgraphs produced in UserBase (this is already in a function there)
  2. Adding these weights to the subgraph network nodes.
  3. Adding an exogenous rate to the "E" equation of MasterEquationModelEnsemble
tapios commented 4 years ago

Yes, I wanted to add the exogenous term back in on the Overleaf (and will do so later today). I suggest we leave it very simple: Just add an exogenous rate eta to the master equations for S and E on the Overleaf, then estimate this eta in the DA process. One could multiply this eta by the number of incident edges that originate outside the network, or multiply it by the global infection rate outside the network, but to keep it simple, I suggest we leave it as a constant rate for now.