We currently compute yearly influenza underreporting by dividing official_cdc_incidence by yearly_positive_tests. But we might be able to get a more fine-grained underreporting factor by also incorporating the share of positive tests. E.g., if at the beginning of a new influenza pandemic, overall positive tests only increase slowly, but positive testing share rises rapidly, there is likely outsized underreporting for a short time period. We currently do not incorporate this information.
Then again, it might be too much conceptual work to integrate this into our current model. I will flag this with Dan at our next group meeting.
https://github.com/naobservatory/p2ra/blob/00ec297987190b3260e25ba3f4ea36905164a648/pathogens/influenza.py#L139-L186
We currently compute yearly influenza underreporting by dividing
official_cdc_incidence
byyearly_positive_tests
. But we might be able to get a more fine-grained underreporting factor by also incorporating the share of positive tests. E.g., if at the beginning of a new influenza pandemic, overall positive tests only increase slowly, but positive testing share rises rapidly, there is likely outsized underreporting for a short time period. We currently do not incorporate this information.Then again, it might be too much conceptual work to integrate this into our current model. I will flag this with Dan at our next group meeting.