Incorporating testing dynamics in the MacPan model

Ben Bolker

2022-06-14

We would like to be able to use the positive/negative tests information in a useful way. Here is one possibility:

That is, each compartment (except ICU/hospital/dead) has a parallel ‘awaiting testing’ compartment. From there, individuals can move back into their original compartment (on the receipt of a negative test), or directly to the recovered compartment (if we assume that all individuals receiving a positive test will self-isolate), or into the next compartment in the epidemiological flow (e.g., if they progress from Exposed to Pre-symptomatic while awaiting test results). All individuals are testing on transition to H, if not before.

This should take care of test lags in a sensible way, although there are a bunch of questions/issues to resolve:

• do test-awaiters differ in behavior, e.g. lower contact?
• what do we assume about test sensitivity/specificity?
• do those who test positive always isolate (i.e. go to R), or are they in yet another compartment (“confirmed”)?
• given a known testing intensity, how do we allocate tests among compartments (e.g. more weight on symptomatic classes, and more to severe infections? How do we model test limitation?)
• how do we extend the transition matrix while keeping the code readable/maintainable?

Integrate with material at private GitHub repo: github.com/wzmli/corona/blob/master/infection.rmd