# COVID-19: Elephant Path Prediction

We now have enough statistics from the USA to draw some useful graphs, so click the Logarithmic options to make the charts comprehensible:

The penciled lines give an eyeballometric fit, but it’s pretty obvious the USA is now dealing with purely exponential infection rates.

Total Cases, which is the patients tested = people already in the medical system, is growing by a factor of ten every eight days. By next weekend, the USA will have one million Total Cases: average it to 112,000 new cases, every day, over the next eight days.

Which may not happen, if only because we may not have the intake / testing / recording capacity for that number of patients and maybe, just maybe, Social Distancing will have an effect. I expect the Total Cases line bend downward slightly during the week, but it won’t be anywhere near horizontal. Obviously, the extrapolation fails completely within the next 24 days, because we lack a factor of 1000 more people to infect.

Total Deaths still equals Total Cases with a delay of fourteen days. By next weekend, the USA will have 10,000 Total Deaths: ramping up to average 1120 new deaths, every day, over the next eight days.

The 9,000 patients who will die in the next week are already in the medical system (because you take about two weeks to die) and, at least in downstate NY, have essentially filled all available hospital beds; they’re getting the best care possible from the medical establishment.

The next 900,000 cases, appearing “suddenly” during the next eight days, have nowhere to go; doubling hospital capacity and converting every flat surface into a mass ward are worthwhile goals, but they’re a linear solution to an exponential problem.

Not every new case becomes a patient, but in the USA we seem to be testing only folks with obvious COVID-19 symptoms, so all the optimistic hospitalization estimates of 10% are off the table and 50% seems more believable. Pick any percentage you like.

Eight days from now, the rate will ramp toward 10,000 deaths per day, to reach 100,000 Total Deaths in sixteen days, again, as an average.

Nearly everybody will survive this pandemic, because the overall death rate seems to be a few percent. For those of us in the Boomer-and-up generations, (theme: Aqualung) well, this may be our contribution to solving the Social Security & Medicare budget problems.

## 11 thoughts on “COVID-19: Elephant Path Prediction”

1. david says:

FWIW, that is NOT “total cases”. That’s “total confirmed cases”, which is a MUCH different number both by meaning and by at least one order of magnitude, maybe two. It’s irresponsible to label is otherwise.

Also, the death rate is only a few percent before we run out of hospital beds. After that, nobody knows, but it’s definitely larger, maybe by an order of magnitude.

“Nearly everybody” is unlikely to survive.

1. Ed says:

As far as I can tell from the numbers, the true “total cases” count lies somewhere between a binary and a decimal order of magnitude higher than the symptomatic-and-thus-tested “total cases”. Absent a well-controlled random sampling of the population, we’ll never know the number, so I’m stuck using the Official Count & Terminology. Maybe “Total Confirmed Cases” would be better, but …

As of a week or two ago, a treatment-by-age summary suggested 1/3 of my demographic will require hospitalization, a third of those go to the ICU, and (assuming everybody dies in the hospital), half of the ICU patients go home in a shoebox. After the ICUs fill up, 10% go home in a shoebox. After the hospitals run out of supplies / space / doctors / nurses, it’s safe to assume anybody in need of hospital-grade care won’t bother making the trip in the first place, so the toll reaches 30%.

Of course, that’s just for my demographic. However, should a kid have a motorcycle collision and need ICU-grade care, they won’t get it, either. Ditto for “elective” surgeries like pacemakers & joint replacements.

Society will definitely be different in five years, but most of the same people will be around …

1. david says:

Yep. I’m younger than you, but for various reasons, my odds are no better. We sure are living in Interesting Times.

2. tantris says:

I’ve been practicing social distancing before it was cool…
And I assume, you’d rather take some appliance apart than spend the evening at bingo-night. Age is one statistical factor, but so is lifestyle. Bicycling & engineering-related activities are safe in my book. And if you can cook, bake, fix things – in short “homestead” – you’re in a different demographic.
Tonight, I modified a coffee mill to make a passable espresso grind (Hipster version of homesteading..). Because the one local supermarket with a decent grinder stopped bulk coffee for reasons of covid-hygiene.
Hmm, I usually treat my coffee with boiling water or steam. But, at least they mean well.

1. Ed says:

I just knew being a dweller in the cellar was a Good Idea™!

Mary planned new garden fences for this year, we just finished putting them up, and plan to be around for harvest season, too.

1. Absent any other information… the amount that you ride your bike (ok, ‘bent) implies “better than median” fitness (represented by VO2 uptake). So, I’m thinking your odds are pretty good. Add to it a high education level, which is a pretty good predictor of not currently smoking, and the odds get better still. Bayesian, but better than nothing. :-) Finally, if you have a borderline case, I’m betting you’ll take the dryer cord off, score a big diode (maybe from the rectifier in the Subaru alternator) and electrolysize some water into O2 and vent the H2 (outside!).

1. Ed says:

Man, this is definitely worse than recommending my better points to Santa Claus.

AFAICT, the goal is to punt an (inevitable) infection at least three months down the calendar, when the surge of medical supplies should meet a declining number of new cases. By then, the medics may have gotten enough sleep to recover their collective sense of humor.

Compulsive hand washing in full effect …

Thanks for the good words; yeah, we are that kind of people.

3. D Schultz says:

How much of that is infection rate as opposed to testing rate? Testing has lagged so badly that we are still just catching up. When I plot the data from https://covidtracking.com/data/state/texas/#history

I get two roughly straight lines of similar slope for positive tests and total tests. (log scale of course)

1. Ed says:

In New York State, you needed a physician’s order to get a test, but they didn’t see anybody without symptoms, so, to a very good first approximation, the “total tested” number equaled the “total infected” number.

They’re now also testing anybody who’s been in “proximate contact” with someone who tested positive and the latest numbers suggest a 40% positive result. I’m not sure where the tracking site is getting total tests or negative tests, as I’ve seen only positive results from the official NYS sites.

Plenty of numbers, just not quite enough!

1. Jason Doege says:

In Austin, they are only testing people with sufficient symptoms, so under-testing. Got a family member in the house with a pending test, right now. 103 fever with a cough and if they didn’t have an albuterol prescription they would not have been tested.

2. RCPete says:

Not seeing numbers like that from Oregon Health Authority. Seems to be dependent on circumstances.

Oregon was stuck with the CDC guidelines* until about a week ago, but there now seem to be private tests in action. As of yesterday, the total hit rate for us was 4.8% positive. The urban hot-spots in and near Portland are running 10% positive. This is offset by the really rural counties running lower. (One county east of us has 5000 total human population and a 0% positive. No testing for coyotes and wolves, though.)

(*) The local authorities said a reagent they needed was only available from the CDC until recently. Still not much testing in our county, running 4.2% positive.