Data basis updated Feb. 4, 2020 at 12:00 p.m. ET
I have been looking for a quick and easy chart to track the Corona Virus 2020. I found a very good and visually appealing display which seemed well researched, with matching numbers presented in the media, and which was displayed on a high reputation website, the Wall Street Journal.
https://www.wsj.com/graphics/coronavirus/ Unfortunately, the site asks you to pay and sign in if you follow this link. For a while, that could be prevented by searching for this site on Google and following the link from Google, this has been fixed in the meantime, meaning that I will stop using this site. For future updates, I will have to take the numbers from the World Health Organisation, which I probably should have done from the start. But anyways.
This presentation is very neat, but in order to get a better grasp of the development, I was looking for a more graphic display, something which would show not only the magnitude of infection at a given time, but also the rate of increase. I was looking for a graph. I wanted to see, does the virus spread exponentially, or are there some signs that the disease gets under control?
Given that Wall Street Journal, withand , had provided a wonderful data basis, I decided to create my own graph based on that, instead of looking for something which might have been created by others.
So here I have several graphics which allow to track the corona virus 2020, graphs to show the development of the epidemic and the rate of infection. As the absolute numbers are very easy to find on the source website, I decided to focus on the form of the curves which I was about to create.
Graph 1 shows the very steep acceleration in new infections. The chart looks very flat until 20. January and then rises seemingly exponentially.
The curve for fatalities is so low in comparison, that a qualitative examination would require a second axis on a different scale, see Graph 2.
As soon as we apply a separately scaled axis for fatalities, adjusted to graphically overlap at the same end point, the curves show a very tight fit, with very similar behaviour.
In order to better understand the nature of the increase, I have decided to put both curves on a logarithmic scale, see Graph 3.
Naturally, this curve looks very different from the previous ones. What I have created here is a semi-logarithmic plot. The x-axis remains linear in order to represent the lapse of time, while the y-axis are both logarithmic, showing a factor of ten between each point on the y-axis. The logarithmic function is the inverse function of the exponential function. So an exponential curve would show as a straight line on this semi logarithmic scale.
Looking at the infection curve for example, it is not straight. It becomes flat in the middle, and then rises steeply, and shows different ascension rates at different times. A simple exponential curve can be put from the starting point to the current end point, to show the difference, see Graph 4.
Graph 4 shows the gap between a simple exponential function, as a dashed line, and the actual confirmed infections development. We see a good match at the start and at the end of the curve. This is intentional, as I have tried to find a function which would have the same start and end point as the actual development to date. The reference function displayed here is 1.164^(day number) – where the 1st of December is day number one.
The gap might be explained by delay in registering and confirming infections. The steep rise appears a few days after the first fatality, as seen in Graph 3. Potentially the infection gained public awareness and people would search medical advice in a more structured way after learning about a very dangerous illness. In other words, maybe the actual development in infections is much closer to a true exponential development than the confirmed infections show.
Extrapolating the reference curve would show about 65.000 infections by next week, 11. February. Every new data point will give us more information about the actual rate of spreading of this infection.
The same can be done to the fatalities chart. We see an S-Curve, compared to the exponential reference curve. This could optimistically indicate that the rate of fatalities is going to be less than exponential. The function of the reference curve in this chart is 1.3^(day number) with 11th of January as day 1.
I do plan to update this page next week, as I find the time.
How bad is the corona virus? How does corona compare to other infections?
- It is much worse than SARS 2002/2003, which had a total of 8.098 infections wordwide. The current Corona virus has already infected a multiple of that, and is still in a phase of roughly exponential rise.
- It has not yet reached the magnitude of e.g. Influenza in Germany 2017/2018, a seemingly “normal” infection, with 350.000 infections and an estimated death toll of 25.100 in Germany alone.
- Mortality rate has not been established yet, as the majority of those infected is still in a state of acute infection.
At this point, it is not known how large the infection base and death toll will be for this epidemic.