By now (June 20, 2020) we have the data of the first 26 weeks of the coronavirus pandemic. I show the weekly numbers of cumulative coronavirus death counts with curves in a figure (Fig. 1). I have curves for six jurisdictions (PRC, USA, UK, F, NL, BR), a seventh for the world and an eighth artificial one (in green) for the model of algorithm 0 (with an R0 value of 2.80 per week). On the right of Fig. 1, curve colors are linked to jurisdictions.
What I have done is focus on what is considered the D-compartment in a SIRD compartment model. SIRD is a SIR model with an extra D-compartment for the deceased. This means that I assume that algorithm 0, when suitably parametrized, can provide useful descriptions of coronavirus cumulative D-curves.
However, Fig. 1 shows that the pandemic has varying start-times and varying curve-forms in different jurisdictions. The task of the modeler is to emulate these curves in compartments/phases with mechanisms that can explain what is happening. My assumption is that this is useful for the first six weeks of epidemics, new in a jurisdiction.
I hypothesize that the coronavirus pandemic has, when it is still new to a jurisdiction and hardly noticed, a reproduction number of 2.80 per week. During the first six weeks the influence of the changes in the compartmental volumes are ignored, as these have vanishing percentages and useful testing and registration are still underdeveloped. The model leads to 269 deaths in week 7, which I consider insufficient to take notice of the effects of I- and R- compartment dynamics.
The idea behind this hypothesis is that by week 6 the legislators become ready to consider measures that help check the disease’s spreading and that by week 7 the population becomes sensitive such measures.
And yes, when I focus on showing what happens with the D-compartment curves of before they reach a total of 1000 (in Fig. 2), I would defend that my hypothesis is not outlandish.
