One month global COVID-19 case load projection with and without Chinese data
The data for this post is taken from https://www.worldometers.info/coronavirus/ retrieved March 16 at 4:00 pm EST.
This graph doesn't address deaths, just the identified cases from Feb 19 to March 16, with projections out to April 16. The projections are all third-order quadratic fits, and the tightness of the fits is given by the R-squared value for each curve. The dividing line between actual and projected cases is the thick line at today's date.
I had previously been plotting both second- and third-order fits, but the current data is diverging further from the second-order fit, resulting in a steadily dropping accuracy. As a result I don't think the second-order fit is useful any more, at least for this dataset. The third-order fits are remarkably tight.
It's clear from the shape of their green line that China has a handle on their spread. To segregate that effect, I plotted the Chinese data separately, and graphed the world data both with and without the Chinese data included. It's evident that the global spread is now being driven by non-Chinese cases. At the moment the major contributors are Italy, Iran, Spain, S. Korea, Germany, France and the USA.
In one month, if social distancing is not effective, the world could have close to one and a half million cases, with almost all of them (1.4 million) outside China.
As of today, the "Total Case" CFR is 4.2%, and the "Closed Case" CFR is 8.3%, based on the reported data for active cases, recovered cases and deaths from the source listed above. It's still early days, of course, but the fatality ratios based on reported data are still climbing.
We're walking (or running?) deeper into the woods