Monte carlo simulations in CARA
Update March 2021: @mrognlie has implemented the MC code for the model in a dedicated branch @nimounet and @pelson agreed to take over this issue.
The parameters mentioned below are no longer valid. See CARA Report for details
Hello. Hope you @all don't mind that I brainstorm on a major implementation proposal. As I mentioned beforehand, the ER parameter is weakest link in the model in terms of uncertainties. A way to treat those uncertainties is to perform monte carlo simulations ranging the ER values for each expiration activity type.
I would like to open this to discussion and have your opinion. Do you have experience in implementing monte carlo in python? I also have a colleague that is willing to help if needed.
Statistical values
Values for infection dose (coefficient_of_infectivity
):
300-1000 viral copies (Mean= 650 , SD=350) ,i.e.
coefficient_of_infectivity
of 0.033-0.001 (Mean=0.00216, SD=0.002166ref: https://www.mdpi.com/1660-4601/17/21/8114/htm
Here we assume that D50 is in the range of 100–1000, and by taking the geometric mean, we arrive at D50 = 316 viral copies. This corresponds to a virus titer of about 220 plaque forming units (PFU≃0.7⋅D50, with PFU being a measure of concentration in virology)
Viral load (viral_load_in_sputum
):
Mean= 7 log10 , SD= 0.71 log10 RNA copies / mL
Aerosol count (aerosols
):
values in mL / m3
'Breathing': AVG=2*10^-3; SD=5*10^-3,
'Whispering': AVG=5*10^-3; SD=3*10^-3,
'Talking': AVG=9*10^-3; SD=2*10^-3,
'Unmodulated Vocalization': AVG=6*10^-2; SD=2*10^-2,
'Superspreading event': #Fixed value,