Closed eidelen closed 2 years ago
The parameter infectious_rate
is one which should be calibrated to your data. The original value was calibrated to the UK for the pre-March 2020 lockdown and should give a doubling time of about 3 days (leading to half the population infected on day 25 in your runs). Clearly there have been many measures put in place since then which lower R and should be included before calibrating infectious_rate
, such as social distancing measures (e.g. set relative_transmission_occupation
and relative_transmission_random
to about 0.7) and self-isolation upon symptoms (self_quarantine_fraction
=0.7). There will also be immunity built from vaccinations (look at the examples) and natural immunity from infection. The parameter sd_infectiousness_multiplier=1.4
is related to super-spreading and is calibrated to give a k=0.5 with the original set of parameters, so probably should not be changed.
Hi, First of all thank you for making your project available to everybody. We are trying to adapt OpenABM for the area of Zürich, Swizterland. We adjusted many input parameters like population size, population age distribution, and so on. However, we kept your calibrated values for
infectious_rate=5.8
andsd_infectiousness_multiplier=1.4
. Running a simple simulation without countermeasures shows that half of the population is infected after 25 days. That was obviously not the case. Actually we don't understand how the calibration step exactly works (why doubling time of deaths?). Even more fundamentally, we do not understand the requirement for a calibration in a agent based model. Shouldn't properties likeinfectious_rate
be the output of such a simulation? Best regards, Adrian