Day 34: Don’t Believe Press Summaries of Data Because There is Really No Drastic Change

• COVID-19 Report As of Saturday, April 18, 2020 GA DPH 7PM
• Consecutive day #2 of decreasing growth (14 needed) ▼
• Confirmed COVID-19 Cases: 17,841 (+647) ▼
  • Daily Growth Rate: 3.7% (-1.1%) ▲
  • Weekly Growth Rate: 5.5% (-0.4%) ▲
  • % of Population Affected: 0.1702% (+0.0061) ▼
  • Hospitalized: 3,447 (+123) ▲
    • Beds Available: 8,322 ✓
  • Deceased 677 (+27) ▼
    • Morbidity Rate: 3.79% (+0.02%) ✓
• Estimated Numbers
  • Double Number of cases every 21 days (was 27) ▲
  • Total Contagious: 142,728 (+6,608) ▼
  • Contagious, No Symptoms: 139,281 (+5,503) ▼
  • Contagious Ratio: 13 in every 1,000 Georgians (+0) ▼
• ▼= getting better; ▲= getting worse; ✓ = same

As you can read in the title of this post, you have to look at the numbers and not what the mainstream media tells you how to feel about the numbers. I read today is a totally incredible story in the AJC today that the death rate has already peaked, that there were just 5 deaths in the last day (changed to 9 in the story now), and that things are so great that we should think about moving to Phase II. REMINDER: We can’t move to Phase II until we have 14 days of continuous decline in the number of new cases. We are at 2 of 14. The original headline was BREAKING: Ga. may have already reached peak of coronavirus deaths, model suggests. Don’t bother to look for it, because you won’t find it online now. The link to the story found here, now takes you to a much more tame story here. Unless you are quick, those stories might be gone by the time you click. They all reference “models” that show we have allegedly peaked! You can find the model they refer to here. This model does not predict the infected, just the hospital load and deaths. Both predictions have been lowered. I never took the time to see how accurate their predictions were because the 95% confidence interval included a very broad range. For those unfamiliar with the slang, 95% confidence interval SOUNDS like they are almost 100% confident in the prediction because 95% is almost 100%. However, the confidence is only that the actual value will be somewhere in the range presented. As my Cousin Joe Miller used to say, there are lies, then there are damned lies, and then there are statistics! For example, their prediction for deaths on 4/18/20 is 40 with a range anywhere from 1 to 181. What we actually had was 27 deaths (not 5 or 9). Yes, 27 deaths was within the confidence range, but every day since March 18, 2020 has been within range of 1 to 181. Thats kind of like flipping a coin and predicting that the result will be heads OR tails. I bet you will be right 100% of the time. I’m not cracking on their model, but reporters and politicians can’t scroll their mouse around a pretty interactive graph and then declare victory. You have to interpret what you are looking at. The model creators aren’t lying–they are just using statistics.

All the politicians are using this model to make decisions. I can tell you that the model is VERY pretty and fun and interactive and therefore very compelling. But it is just a best guess from some people playing with some numbers. Predictions are not a contract with the COVID-19 virus. The best thing I discern from the model is that it appears that in Georgia we will have enough hospital beds and ICU beds to handle the sick people. So if you were envisioning a hospital visit like the first episode of The Walking Dead, you can put that out of your mind. Busy? Yes. End of the World busy? No. That will be good news when it stops being a prediction and becomes a reality.

The actual good news is that most of the metrics are still in the green. See above. The red numbers are mostly due to the suspicious drop yesterday followed by a more reasonable set of numbers today. Hospitalizations went up almost twice as high as the day before, however. Fig 1 shows the continuing decline of the average rate of growth. Fig 2 shows the remaining numbers. I promise that I’m not cheating, but my corrected predicted number (green line) has been pretty much on track. I’m adjusting daily, but it is still a prediction. I might remove the orange (exponential growth) and yellow (average growth) from the graph in the future, just to make the corrected prediction stand out. As you can see, however, the predicted number continues to go up. The more people confirmed, the more infected, and the more infected, the more people that could die. Today’s morbidity rate is 3.79%.

Update: I forgot to update my predictions before publishing the first time. Fig 2, above, contains my prediction for Sunday. I think in the future that I will predict 7 days in advance every day instead of a week at a time.