Is anyone planning to add a feature to EpiNow2 and epinowcast that would account for imported cases? That would be super useful. Thanks!
Edit: our team has been providing real-time modeling support for a cholera epidemic for some weeks now. Initially, it seemed that imported cases were driving local transmission, which led us to use EpiEstim because it can account for imported cases. We later found out that the imported cases were likely not contributing much to local transmission due to quick isolation. In any case, this feature would be nice for future outbreaks in which imported cases are important.
Hey Kelly, thanks for bringing this up! I suppose you mean a feature to support nowcasting of growth rates/reproduction numbers under imported cases?
In terms of modeling, I think this should be rather straightforward to achieve, but would require a bit of implementation work. Unless there are loads of imported cases, one would have to assume that they have the same reporting delay distribution as internal cases (which may be questionable depending on control/tracing measures). Imported cases would then have to be smoothed using a separate non-parametric smoothing prior, and integrated into the renewal model for internal cases accordingly. Here of course the tricky part is that if cases were infected abroad, their generation time distribution is left-truncated to the time of entry. If you know the date of entry, it could also be modeled, but this makes things more complex of course…
What do other people think?
Yes, that is what I mean. Would it be possible to specify a different reporting delay distribution for imported vs local cases (assuming you have the information to estimate it)?
Sorry for the late reply, yes in principle nothing speaks against that but I wonder how realistic it is that you have enough cases to estimate this…