search

rajeshrinet / pyross

PyRoss: inference, forecasts, and optimised control of epidemiological models in Python. github.com/rajeshrinet/pyross
https://pyross.readthedocs.io
MIT License
169 stars 57 forks source link

Exposure latency #2

Closed Chinmaya-Kausik closed 4 years ago

Chinmaya-Kausik commented 4 years ago

I'm not sure, but it seems that adding exposure latency, and hence having an SE(Ia Ir)R model, would help. I'm asking here because this is one of the few models accounting for age-based social contact.

ronojoy commented 4 years ago

@Chinmaya-Kausik yes, we could add exposure latency, and have an E class, but that would only increase all the estimates we have made. The SIR compartmentalisation, with two types of infectives (asymptomatic and symptomatic) will give a lower bound, which is the "best case" scenario we have described. Note also, that we set asymptomatic cases to zero, since we have no idea how many such cases there are. This further reduces the bound, since there are asymptomatic cases, possibly much larger than the confirmed symptomatic cases.

Chinmaya-Kausik commented 4 years ago

Indeed, that is true. I was merely interested in what would happen if we try to increase accuracy. Also, you have probably seen this already (and it is of course probably not indicative of anything due to low sample sizes), but here is an estimate for asymptomatic cases on the Diamond Princess. It may be somewhat illuminating to run it with alpha from here.

rajeshrinet commented 4 years ago

@Chinmaya-Kausik thanks for bringing it to our notice.