A library of published compartmental epidemic models, and classes to represent demographic structure, non-pharmaceutical interventions, and vaccination regimes, to compose epidemic scenarios.
This issue is to flag that the Vacamole model is not correctly set up for rate interventions (and for time dependence).
Specifically, the passing of model arguments/parameters as a proportion reduction in the value of other arguments creates challenges in applying rate interventions and time dependence. For example, the argument susc_reduction_vax specifies how the transmissibility of fully vaccinated individuals is reduced, as transmissibility_vax = transmissibility * (1 - susc_reduction_vax). Since the subtraction step occurs before the model is parameterised, any change in susc_reduction_vax is not passed through to the ODE solver.
This could be fixed by restricting rate interventions to the primary model rates, such as transmissibility, infectiousness, or the recovery rate, thus allowing proportional decreases due to vaccination to work as before.
Alternatively, the vaccination-modified parameters could be passed directly, e.g. as transmissibility_vax, rather than being a function of transmissibilty and susc_reduction_vax.
Finally this throws up a more general question of which features are required in which models, and how to make existing models compatible with these features.
This issue is to flag that the Vacamole model is not correctly set up for rate interventions (and for time dependence).
Specifically, the passing of model arguments/parameters as a proportion reduction in the value of other arguments creates challenges in applying rate interventions and time dependence. For example, the argument
susc_reduction_vax
specifies how the transmissibility of fully vaccinated individuals is reduced, astransmissibility_vax = transmissibility * (1 - susc_reduction_vax)
. Since the subtraction step occurs before the model is parameterised, any change insusc_reduction_vax
is not passed through to the ODE solver.This could be fixed by restricting rate interventions to the primary model rates, such as transmissibility, infectiousness, or the recovery rate, thus allowing proportional decreases due to vaccination to work as before.
Alternatively, the vaccination-modified parameters could be passed directly, e.g. as
transmissibility_vax
, rather than being a function oftransmissibilty
andsusc_reduction_vax
.Finally this throws up a more general question of which features are required in which models, and how to make existing models compatible with these features.