TakeItAndRun
commented
4 years ago Sorry for the bold type. This was not intended.
Closed TakeItAndRun closed 4 years ago
TakeItAndRun
commented
4 years ago Sorry for the bold type. This was not intended.
TakeItAndRun
commented
4 years ago Sorry I posted this in the wrong GitHub. It was meant as a comment to:
by
https://www.mongodb.com/blog/post/tracking-coronavirus-news-with-mongodb-charts
Please close and ignore.
Thank you for your work. You are one of the first (that I know of) to plot the data in a logarithmic form for every country.
I'm very much interested at doing exponential regression of the case numbers. On a log plot these appear as strait lines. I have done this here for the case numbers for Germany, where I live. (Data from the GitHub of the Coronavirus COVID-19 Global Cases by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University (JHU) ).
Actual, I did not do an exponential regression here, but put a line on the plot by eyeballing it. (I have studied atomic physics at KSU,USA , where I did a lot of data analyzing.)
On this plot I multiplied the data by 15% to estimate the number of people needing hospitalization (They mostly need oxygen.) by 5% to estimate the number of people that needing an ICU bed. by 2% to estimate the case fatality rate
These estimates are numbers from China. But I don't have the original source at the moment. Here these numbers a quoted: https://www.statnews.com/2020/03/10/simple-math-alarming-answers-covid-19/
The number of hospital beds in Germany I took from: https://miro.medium.com/max/939/1*ifPqpODBwlZaa_c94Onyzw.png The number of ICU beds in Germany I took from: https://www.businessinsider.de/wissenschaft/gesundheit/so-bereiten-sich-die-kliniken-auf-den-ansturm-von-corona-patienten-vor/
I multiplied the number of beds by 0.5 to account for that not all beds are available for corona cases, they are filled with peoples with different ailments. Please be aware that the number of beds are only a quick estimate of mine. Further study is needed here.
I also plotted the number of actual death. (IMHO, the number of death at the start of an epidemic is somewhat imprecise. France, with similar case numbers as Germany has today a death rate of about 2.35%)
The prediction I make from this plot (5% of total cases needing an ICU bed) is that given no change in the transmission rate, the number of ICU beds in Germany will be filled by Sunday, March 29, 2020.
Looking at the number of actual death, and multiplying them by 2 (as the death rate of people in the ICU is about 50%) I would estimate the number of ICU beds in Germany will be filled by Monday, April 6, 2020.
If you want to play around with the numbers yourself, I include the excel file (generated with LibreOffice Calc). I changed the extension from .xls to .txt to upload it to GitHub. Please reverse.
Corona Cases ans Hospital Beds for GERMANY.txt
Today my government announced new measures of social distancing: starting Wednesday all nonessential business close, etc, etc... Lets hope that the doubling time will decrease in the near future.
Such a detailed analysis, for one given country, may well be beyond the scope of your chart.
But an exponential regression for the case data and the calculation of the doubling time for each country would be well worth it. One problem is how to select the linear range (in the log plot) to use for the exponential regression. The last strait section should be used to estimate the (current) doubling time. I don't know how to do this automatically.
To make your log plot more readable I would suggest to group countries with about the same range of cases into one chart and limiting the range of the x-axis and/or y-axis. In the very beginning of an epidemic the increase of cases is in single digits and somewhat random. It is only with the establishment of testing and the accumulation of some statistics the numbers become meaningful to look at for e.g. doubling rates.