COVID-19 March 19, 2020

A couple of important changes with this update. First, going forward, I will provide an analysis using only the prior day's data. I've had to make this change, because the CDC releases info at noon (EDT) and those data are woefully incomplete and behind compared to the reporting from the individual state health departments. As soon as the day's numbers from the CDC hit the street, they are already obsolete. For example, yesterday's numbers ended up at 8,058 cases, whereas the CDC report from yesterday showed only 7,038.

Second, I am now using wikipedia as a data source, which in turn posts data directly from the state health departments. See here and here. These data are continually updated in near real-time, and are verifiable by going straight to the state health department websites that are given as references for the data.

So, today's update is really a reanalysis of yesterday, but using more accurate and complete data. There's no point in looking at what the CDC will say later today, because the data are still pouring in, and will be continuing to do so until late until the evening. I'll look at those final numbers and report on them tomorrow.

Below is the analysis based on data valid for yesterday (click on the image for hi-res).

As I indicated above, the actual number of cases (8,058) was far above what the CDC had reported at noon (7,038). Including this more accurate data drives the short-term trend line (orange) closer to the long-term trend (blue). I can't be certain, but I strongly suspect that effort to increase testing is starting to show. Still, the number of suspected cases far outweighs the current testing capabilities. As capabilities increase, I expect the short-term trend to approach and then actually surpass the long-term trend. It seems very likely that the slight dip over the last few days was due to systematic bias in testing, and not a true deceleration of the exponential growth.

The short-term trend line is predicting 10,129 cases by the end of the day. That is likely to be low compared to reality and indicates that the short-term trend is indeed suffering from systematic data bias. Using either the short- or long-term trend, the data predicts that we will hit 100,000 cases around March 26, plus or minus a day. This is 10x the current number. If things continue, we will hit 1,000,000 a week later. The 100,000 number is nearly a sure bet. There's nothing we can do about it because the people that will be diagnosed are already infected. If significant actions are taken, there's a chance that we can have an impact on the 1,000,000 two weeks from now. It will be interesting to see if the patchwork of half-assed local and state policies will show any effect; it's going to take more than a week to assess that.

Flattening the curve is a popular meme right now. Below is an analysis of the progress, or lack thereof, in doing so (click on image for hi-res). The graph shows the daily change in new cases. For algebra buffs, each data point shows the slope of the curve in the graph above for each day. For calculus buffs, it's the derivative of the curve above. If the number of cases was to remain the same as the day before, the line connecting two data points in the above graph would be horizontal. Horizontal lines have a slope of zero. Clearly, that's not the case. The line is not horizontal. It has a positive (and large) slope. Right now, the number of cases each day is increasing exponentially, and the slope of the line is getting larger and larger, which is why the line in the graph below is going up. In other words, the curve is not flattening at all. A flattening of the curve would mean that the data in the graph below would start to fall and head toward zero. A value of zero, however, indicates only that the number of cases isn't increasing or decreasing each day. Ultimately, we want the number of new cases each day to be lower than the day before. That would imply a negative value (below zero) for rate of change. We've got a loooonnnnggggg way to go before that happens.

The fate of all these curves depends exclusively on the collective actions of individuals. We must all do our part. Do not go out in public unless it is absolutely necessary. The CDC recommends no groups larger than 10. This is inadequate. Groups no larger than 1 should be avoided. In other words, no groups!

I continue to receive information from my doctor and nurse friends working at local hospitals. They are already running out of beds, supplies, personal protection equipment, and everything else that is necessary to treat the infected. We're only at 10,000. Next week it will be 10x worse, and the week after will be close to 100x worse. The system will likely have exceeded its breaking capacity. Nobody knows what will happen then. There is no plan. In Italy, patients were left to die. It's hard to imagine a different outcome here. Each of our individual actions can make a difference. DO IT.

EDIT:  CDC Website now says the following:
"In the event of a discrepancy between CDC cases and cases reported by state and local public health officials, data reported by states should be considered the most up to date."  The data above are from the state and local public health officials, as indicated.