Earlier, Mr Node gave a few tips to read his tableaux. Here he shows one in action. As an example, he again uses the version instantiated with data for the Netherlands. But now he also shows the row on the right, of control data for the program that creates the image. Here we go: first the row of parameters, then Graph 1, then Graph 2 and, finally, Graphs 3 & 4.
The parameters
First the jurisdiction (in this case: the Netherlands), which is chosen from the six built-in options.
Then root-shft. It is likely that before the first deaths are recognized and registered as COVID deaths, there have been previous infections and deaths that have not been recognized. This means that for the quality of the model of the COVID deaths D (0), a starting point can be chosen that precedes the first deaths registered as such. With root-shft the user can choose, with week 0, which is fixed on December 29, 2019 as an anchor. For the Netherlands, Mr Node set the beginning of the epidemic on week six (the week after January 31) instead of week 9 in which the first deaths were registered. This is in line with what is now believed that happened actually.
Then xmax, the number of weeks for which observations are available. Week 0 ends on December 29, when there are no registered deaths. Mister Node faithfully picks up the new numbers every Saturday and enters them in the tiny database that comes with the program. At the time of writing, he is at week 29 (ending July 17).
Then r-zero or R0. That is the reproduction number found by applying it in the standard model. Mister Node experiments with this until he finds a line that initially coincides with the observations and at a certain moment distances itself from them. The working method is based on the idea that when a jurisdiction is attacked by a new pandemic, it will take some time before effective measures can be taken.
Then d-Gap and Incutime. When we have registered figures for who are infected and who have died, we can deduce everything (Node thinks) necessary when we have an idea of how long it takes for someone who is infected to get sick and how long the sickbed is for whom die. There are, of course, all kinds of individual differences, but as long as he has not yet started to build micro models, one can work with assumptions about average duration. Mister Node currently opts for a 1 week incubation period and a 3 week course of the disease that either ends in recovery or in death.
The three gray parameters are the buttons that activate the program. And the four green values located below them will be discussed later.
Graph 1: Calibreren voor fase 1
Graph 1 shows an enlargement of the first phase. That takes place in the Netherlands between weeks 6 and 14, says Mr Node. We then have a period of 9 weeks (!) In which reality corresponds to the basic model of algorithm 0, when applied with an R0 value of 2.65. In that week the two lines clearly diverge. It is striking that the three lines initially show all three of the dreaded curve upwards, but that the D (0) line changes into a straight line for two weeks and then turns to show a downward trend after week 17. This is small-scale work that Mr. Node shows separately because the details are lost when showing the same information on a wider scale. It should be noted that a number of key figures are included in graph 1 because they indicate which scales are involved. For the whole picture, we need the space for just under 52,000 cases and 29 weeks.
Graph 2: a different scale – more is different
That image is provided by graph 2. It contains the complete trajectories of the registered (and thus observed) data of I and D (o), both black. It can be noted how D (m) continues to follow its curve upwards. On the way to week 18 it disappears out of sight (greater than 52,000) and Node previously pointed out that at week 23 the line even exceed the number of the total Dutch population.
The red lines R (m) and S (m) are based on the assumption that the mean incubation period is 1 week and the mean disease-course length is 3 weeks. Based on these values (and the numbers of deaths observed), estimates can be made of the number of cases leading to a cure and of the weekly scores of the numbers of sick people. These lines are red, because they have to be checked against the observed facts in order to be trusted. About that later.
The interpretation of the lines shown in the graph seems encouraging when we look at the curves and arrows we can recognize in them. But perhaps the scale is too large for this.
Graphs 3 & 4: daily rates (or rather: weekly figures)
Finally, Mr Node looks at graphs 3 and 4. These relate to weekly figures. There have been weeks when seven and a half thousand new infections were registered – that’s more than a thousand a day. For mortalities, the maximum is around 1,000 per week (or 140 per day). The maxima fall both in week 15. The maximum number of patients is a bit behind (week 18).
Here, too, the curves and arrows are clearly recognizable, and it seems that around week 28 a tipping point is emerging that could announce a second wave.
For the time being, it appears that whether this will occur depends on the extent to which the measures are respected. A morbid side of the matter is that particularly low-risk adolescents who are (for good reason) still biologically equipped with preferences for ignoring rules and looking for dare-all risks, can act as intermediary carriers as a result of their mass meetings. In this sense, the idea that Node read somewhere, the “Homo Homini Virus” idea is by no means far-fetched.