I was pleased to see an absolutely crystal-clear indication based on my projections, that Washington State has dramatically reduced the peak impact of COVID-19, and it’s unambiguous that at local levels, distancing can have a huge impact.
The earliest case diagnosis (21 January, which seems like another century ago) of COVID-19 and relatively clear communication and reaction at the time seems to have created the appropriate distancing reaction fairly quickly, and with high confidence by communities in health and governmental warnings and information. At the time, I was myself in the Seattle area weekly, keeping an eye on the news in my hotel each evening, and online.
By contrast, New York’s first case/diagnosis, 40 days later, was nested inside an envelope of fatal miscommunication at the national level, and I believe made it very difficult to gain community consent for distancing, quarantine, and shutdown. I’m sure there are other factors, but that’s the most stark one to me.
You can also see from this chart what underlies the discussion of a national “Peak” to COVID-19. The darkest, most overlapping among all the curves is around 29th April — 3rd of May. Thats the day when things will be the worst nationally, and then slowly pull back.
(Apologies for it being an eye-chart, and the somewhat redundant color scheme from Excel. I don’t know how to mass change curve colors. it was relatively complex to create in Excel but I thought important to get out in any form. I checked the data several times. I don’t believe I’ve made any systematic errors. As I refresh it, I’ll work to make the labels clearer.)
The tracking chart for today:
We surpassed the total Military deaths in Iraq with COVID-19 on the 1st of April , and at the national level I don’t see any major signs of slowing in the rise of total fatalities. We also reached another milestone, a peak daily death count of over 1000.
COVID Daily Fatalities Progression
Date |
Fatalities |
4 March |
11 |
17 March |
110 |
1 April |
1049 |
11 April? |
10794 |
I’ve discussed with a few people this daily increase, which is staggering, but will stop. When you hear on the news the “Peak” or other words, when I have said “Peak” in other postings, that’s the day when daily Fatalities have reached a plateau, and should decline as steadily as they rose. I was also including IMHE projections, but I’ve removed them. They simply revised up their early optimistic projection by 16%, to near 95,000 fatalities. Their projection also has roughly double the error to actual data than the one I use. I suspect their forecast will rise quickly, and my forecast will slowly fall and over time they converge, an effect of having similar underlying mathematics.
The line you see here was built by me with what’s called a “Generalized Logistic Function” often used in modeling biological systems. I take prior day data (now from New York Times GitHub repository), and Excel, and fit the formula parameters to actual data, population and Case Rate Fatality assumptions. My curve has a 5-day rolling average forecast error of around +/- 2.5%, and has been hitting daily accuracy peaks of 99%.
I foresee that by April 5th-7th, I’ll understand if my upper-end assumptions are too high (increasing error), or close enough, because those days will be an inflection point in growth based on the speed that Coronavirus spreads. I’m hoping I’m way too high.
Stay Calm. Critical decisions need a clear head.
Stay Sane. Incessant COVID noise from deranged people will derange you.
Stay Home. There’s no safety in numbers, for the time being.
Disclaimer: I’m not a statistician, nor am I an epidemiologist or physician. I do have substantial training in science and mathematics. My assumptions are based on publicly available data, and interpreted without domain-specific expertise to temper my judgements.