A little about our amateur COVID19 model

(Notes by an armchair virologist)

Several people receiving my daily forecasts have asked about the process of estimating the so- called “true cases”. Indeed, how do we know what true COVID19 cases are if testing is partial and random, and none of the available statistics provides us with an answer to a pretty basic question: how many people in a given pool are actually infected? Fortunately, something called “Compartmental Epidemiology Modeling” comes to the rescue and can help even lay modelers like myself.

What follows is a brief (and hopefully not too boring) description of the process.

1. INTRO TO BASIC COMPUTATIONAL EPIDEMIOLOGY (Don’t be alarmed, it’s not that scary).

In devising the forecasting model, I am relying on a basic epidemiology model called SIR (also known as Kermack-McKendrick model).

Briefly, the SIR model suggests that the population is compartmentalized into three groups: Susceptible (“S”) – people who have not yet been infected; Infected (“I”) – people who are currently infected; and Recovered (“R”) – formerly infected people who are now recovered. Together, they represent the entire population, so that:

S + R + I = 100%