Coronavirus

[eevee-fan]

Looking for volunteers for the control group.

Wynn Las Vegas accepts reservations for Memorial Day weekend

 

[Beltsbear]

Good news today, the death rate looks like it is going down, at least on Worldmeters.com.

 

[eevee-fan]

Good news today, the death rate looks like it is going down, at least on Worldmeters.com.

Last 1100's was Apr 5. The decrease is a result of NYC's response to get things under control...

 

[AZRI11]

I strongly argue against the accuracy of this for CA.

LA alone had 600 new cases yesterday.

I live in Rhode Island. Over 200 cases today population is only 10% of L.A. county. And this is the lowest we've had in a while. Northeast corridor is hot. CA has been cold in comparison. All those school trips to Europe. An annual "pilgrimage" for private schools on their breaks in the northeast.

 

[dhanson865]

Here's a tracking site to make these comparisons easy.

COVID Projections Tracker

as far as I can tell it just shows multiple tracks for various projections. I don't see any comparison there like this which uses actual deaths.

compare those two and they are nothing at all alike to me.

 

[davepsilon]

What would you say mathematically drives the change in growth rate? For example, dlogN = f(x). What would you measure for x?

Looking at the google mobility reports I was reminded that the real world has ramps and asymptotes to behavior changes. Naively I would have expected step changes in response to public interventions like stay-at-home orders. But people’s behavior is very complicated and a moving target. Not an answer to your question but just some rambling thoughts that remind me how hard the task is.

You are trying to fit a curve over time through both an effective transmission rate that is changing and an active/resolved ratio that is changing. I think an asymptote on the effective transmission rate is screwing up your model so that recent points are now moving downward slowly at the timescale of disease resolution (up to a month to go from new case to death/recovered) - but that’s just my own guess and I wouldn’t weight it too heavily. Also your curve fit implicitly makes a fixed public policy assumption which is likely to reduce its accuracy over longer time horizons because it seems unlikely to stay that way.

 

[traxila]

Maybe my memory failed, but what I recall reading was that they were viable.

Coronavirus RNA detected on cruise ship 17 days after passengers left: CDC report

My reading of this is that they were not viable.

But I do not blame you. The headlines that I recall for this news item were pretty much all slanted to imply that not only was COVID 19 primed and active after 17 days, but that it had hired veteran mercenaries skilled in lethal hand to hand combat skills as it prepared to steamroll all of humanity one way or another.

 

[sparcs]

Good news today, the death rate looks like it is going down, at least on Worldmeters.com.

data is lumpy. I'm not sure i would draw too many conclusions based on one day.

 

[TheTalkingMule]

Data lags on the weekends, we'll likely see a massive uptick tomorrow.

 

[powertoold]

data is lumpy. I'm not sure i would draw too many conclusions based on one day.

That's true, but this drop is probably 2-3 std deviations from the norm.

 

[SageBrush]

My reading of this is that they were not viable.

An RNA test does not tell you anything about viability. It's like saying you found someone's hair clippings.

 

[EVNow]

Here's my first attempt to show which states are past their peak daily rate of cases (not deaths):

View attachment 536091

Red is bad, green is good. I chose 7 days as the dividing/neutral line because that would be my criteria for calling the peak.

MA has actually been below 10% for 12 days but their reporting of cases is "chunky". For example, its Apr 25 number was 15.9% while the previous day was 0%.

NE and IA percents go in waves above and below 10%, so it's difficult to get a handle on them.

So many states are green because they don’t test enough.

Also, I feel the rate of growth / log graphs hide a lot of information.

If a car is going at 100 mph, when the limit is 40 mph, both growth and absolute numbers are important.

 

[SageBrush]

1. They are covered in fat, so soap breaks up the fat and kills them. Alcohol also kills them. (Not sure why Trump didn't say people should drink more.)

Or eat soap

 

[EVNow]

If you mean ( new_positives / new_tests ) then here's the graph for PA:
View attachment 536203
Looks like it peaked around 10 days ago. I'd ignore the recent wild swings.


We don't even know who's being tested, so I think it would be impossible to come up with a good threshold.

WHO suggests between 2% and 12%, IIRC.

I think anything < 5% is good. Above 10% not so good. Most US states are above 10% - esp the ones with outbreaks.

 

[dhanson865]

Germany: In March 2020 Plug-In Electric Cars Sold Like Never Before

Almost 1 in 10 new passenger cars sold in March in Germany was a plug-in.

March 2020 was an exceptional month in Germany, as the coronavirus lock down significantly affected the automotive market (new passenger car registrations collapsed by 37.7% year-over-year), while the plug-in segment noted record sales (despite the lock down).

The number of new passenger plug-in car registrations amounted to an all-time record (for the third straight time) of 19,755 (up 104% year-over-year) and record market share of 9.18%!

EV subsidies
One of the biggest factors behind the EV sales surge in Germany are more generous subsidies (since February 2020, "retroactively applicable to all vehicles registered after November 4, 2019").

Environmental bonus for new cars:

• BEV under €40,000 gets €6,000 (previously €4,000), while under €65,000 gets €5,000
• PHEVs under €40,000 gets €4,500 (previously €3,000), while under €65,000 gets €3,750

Also, some used cars can get the same subsidies, if: are under 1 year old, under 15,000 km and didn't get a comparable subsidy in the country of the first registration.

So virus reduced sales of cars in general but electric car sales increased both in percentage and in gross numbers.

edit: ah the subsidies pushed people to buy electric and the virus kept things quiet otherwise.

 

[dhanson865]

750,000 tonne potato surplus that has built up because of coronavirus. Belgians urged to eat frites twice a week to deplete coronavirus potato mountain

 

[Oil4AsphaultOnly]

Damn. You have some good ideas. Care to join our Scientific Advisory Board? I can give you a position right next to the psychotic proctologist. The position comes with unlimited free steam cleaning.

Is that before or after the soap gets dropped? As I'm a little concerned that the "position" wouldn't be mutually beneficial. Unless I'll be well compensated for my seat-of-the-pants expertise?

 

[FoverM]

So many states are green because they don’t test enough.

Also, I feel the rate of growth / log graphs hide a lot of information.

If a car is going at 100 mph, when the limit is 40 mph, both growth and absolute numbers are important.

The underlying process is exponential so a log graph is appropriate. A log graph lets you know how things are changing "right now" at any point along the curve. The best example is a stock/index chart. Look at a linear graph of the SP500 since 1980. The disaster in Oct 1987 is barely a blip ... yet I lost 20% of my investments in one day! The Great Depression is even less noticeable and that was, what, a 90% decline.

We're all reading tea leaves here, even the "doctors", because we don't know what's being sampled by all the testing. Look at the prison results:

In four U.S. state prisons, nearly 3,300 inmates test positive for coronavirus -- 96% without symptoms

The "patients" are in a controlled environment. From that article, we don't know what kind of test was performed, but 444/723 (61%) were positive. And 98% of the positives were asymptomatic. Should we trust any of those numbers? I talked with my sister, a microbiologist, about this stuff and she said that the parameters (specificity/sensitivity, etc) on the current tests meant they were "crap" (edited for a family audience!). Even if the tests were "crap", how could they return such high positive rates??? Flipping a coin (i.e. a random test) wouldn't produce such high numbers.

Is there *any* definitive test out there? We've spent 2-4 *trillion* dollars so far on this debacle. Where are the gold standard tests for this damn virus?. I don't care if it requires a scientists chained to electron microscopes all day mapping electron orbitals on individual atoms in the virus's RNA.

 

[AlanSubie4Life]

as far as I can tell it just shows multiple tracks for various projections. I don't see any comparison there like this which uses actual deaths.

View attachment 536244

compare those two and they are nothing at all alike to me.

Just have to select the right one. You can see how the curves have changed over time. Really more of a tool for seeing how the model changes over time. Yes, it's true that you can't concurrently put the upper and lower bounds on the same plot with this link; you can't plot the envelope as in your plot. But for each of the models plotted, it puts in the actual deaths up to the date when the model was discontinued, and then completes the different lines with the various projections of each of those models.

So here's the closest approximation (I've turned off some projections):

It is possible to combine actual death data with the maximum or minimum bounds, but it ends up looking odd - but it's kind o similar to what your plots show...just requires some imagination!

Here is death data up to a given date, which is then completed by the the projection of the upper confidence limit on deaths, for each of several models.

 

[NicoV]

750,000 tonne potato surplus that has built up because of coronavirus. Belgians urged to eat frites twice a week to deplete coronavirus potato mountain

We also have a beer surplus to take care of; and out of solidarity with the French we need to consume more wine and Champagne. With obesity being a mortality factor, no wonder our mortality rate is so high