a device that is generating concepts

1 Comment

COVID-19 (Coronavirus) post-incubation

[October’20 update at the bottom. Hint – really good prediction so far: Israel Covid Dead[+14days] = 0.82% of Confirmed (rsq = 0.9236)].

Those days, when you have to stay (and work from) home, you realize how interconnected our world is – because someone in China distant province eaten something bad, you and all around you (regardless where you are from) stay at home with all kids for weeks or months, gather food, cancel flights, loose savings… in best case, or life in worst. How can we say after all, that this is not a unified world we are leaving in. Enough philosophy – let’s get to the data.

There is an ambiguity about incubation period of COVID-19 and it is defined between 2 and 24 days. Definition of incubation period is “the time elapsed between exposure… and when symptoms and signs are first apparent“. There is however another important index – the duration of disease. Based on the data we can get the duration after exposure till critical period. Assuming “Total Cases” index is captured (on daily basis) after incubation period in most cases and after latent period (the time from infection to infectiousness) in some, we can run correlation between the “Total Cases” [from here: TC] signal and to “Total Dead” [from here: TD] index signal, capturing Rsq while changing time shift between them. The peak of Rsq would expose the Time-to-critical-point of the disease.

Dataset source: Johns Hopkins University Center for Systems Science and Engineering (GitHub; HDX) captured at 16/3/20

Since China has changed on the way the metrics, therefore the dependency between the TC and DC got duality in behavior as we will see, so I have split the data for analysis.

Here is correlation coefficient between TC[day] and DC[day-i] and how correlation looks at peak for all the world except China:

Total Death [after 12 days] ~ 0.248*Total Cases
Total Death [after 6 days] ~ 0.09*Total cases
Total Death [after 0 days] ~ 0.03*Total cases

You can see that as we go towards peak of death index at 12 days after incubation among captured cases, the multiplier becomes scary-high – almost quarter. Really hope that is not the case and just a transitional distortion…

Continue reading