@whelanh1 Thanks. Model is my own work, but having since seen a little more of how other people tend to do it I can see it would be classified as compartmental-type, although with various modifications & additions. It models a process as a flow through states. There is, in fact, ...1/n
@whelanh1 2/n no specifically generated Gompertz. It's a consequence of the process. As far as I can see, the output of a basic compartmental model is Gompertz. The simplest theoretical process is the same as that represented by Gompertz. That's why Gompertz works so well for people. ...
@whelanh1 3/n In practice I find that there are subtle differences to pure Gompertz: you can see it in the data for various locations whose log plots are quite linear but do in have kinks to them.
Modelling lockdown: Within the model is the infection stage, in which each infectious ...
Modelling lockdown: Within the model is the infection stage, in which each infectious ...
@whelanh1 4/n person attempts to infect "x" other people per day (I say attempts, because more become immune as time passes). Lockdown, since it would theoretically reduce contact, is modelled here simply by reducing that "x" during the period in question, then restoring it after.
@whelanh1 5/n The sort of result you get is what you see there for FL. The model log plots I show, btw, are generated from the model output - they are not part of what generates the output. The process generates the function (rather than vice versa). You could do lots of other things ...
@whelanh1 6/n re lockdown. Another example I experiment with is to model the virus being contained before it has spread widely in which case lockdown might shield people by removing them from the 'susceptible' list. Various ideas. But importantly, for me, it has to be a reasonable ...
@whelanh1 7/n underlying process & I try not to add more assumption (& so bias & error) to it than necessary. Model is not reality but I want it to be a reasonable hypothetical versions of it & not just a nice curve fit. Not sure if that all makes sense. Happy to discuss further.
...end.
...end.
@whelanh1 @threadreaderapp unroll
