Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Iduno]

It was looking like it was getting better, but then over 6000 deaths were recorded in one day in the USA of coronavirus recently.

That puts the deaths to known cases at around 5%, which implies that it's just the tip of the iceberg.

That is a reporting anomaly, not an actual 6000 deaths on one day - and the data has tons of asterisks by it because of that.  Although to be fair, we don't really know how many deaths there are due to the disease on a given day.

This also goes to show why you should never change reporting criteria in the middle of a data set - you should instead fork the data sets - continue the old way in parallel with the new way for instance.  Or just stop the old data set and start the new one.

Yeah, but for this example, the old data set was shown to be incorrect, so you'd discontinue it and replace it with a better one.

[Starver]

[Iduno]

https://twitter.com/i/status/1250214461622738946

Also, one about making masks:
https://twitter.com/moby_dickhead/status/1250567120032878592

[misko27]

You know I've mentioned how I'm worried for certain countries... well, I've been following the NY Times Coronavirus map and tracker and see some worrying trends. Russia has surpassed Brazil's (official!) totals, which is concerning as Brazil is the larger country and they previously had a very similar case count per capita. Brazil's case count has raised somewhat more slowly than Russia, but it's deaths have almost doubled from a week ago it seems like, which is really bad because the ratio of deaths-to-official-cases was really very high, indicating a vastly higher case count than is reported.

On the topic of vastly higher case counts than reported is Sweden. Sweden has 12,500 cases, but 1333 deaths. Their ratio of cases to deaths is actually much worse than Brazil's right now, and Brazil was though to have several times the official caseload by scientists there. (The ratio of cases to deaths in Brazil is 6.32%, the ratio in Sweden is 10.63%). That's... mind-blowingly bad actually. Implies either the Swedish Healthcare system is 40% worse than Brazil's (which I don't find likely), that the Swedes have some sort of super-strain which is straight up 250% as deadly as the regular virus (which I really really hope is not the case), or they're dramatically undercounting their totals.

[TamerVirus]

China revised Wuhan's death toll from 2,579 to 3,869
EXACTLY 50%, rounded up 1
such a nice, round percentage...

On the topic of vastly higher case counts than reported is Sweden. Sweden has 12,500 cases, but 1333 deaths. T

Isn't Sweden going the whole "no lockdown, herd immunity" route?

[ChairmanPoo]

You know I've mentioned how I'm worried for certain countries... well, I've been following the NY Times Coronavirus map and tracker and see some worrying trends. Russia has surpassed Brazil's (official!) totals, which is concerning as Brazil is the larger country and they previously had a very similar case count per capita. Brazil's case count has raised somewhat more slowly than Russia, but it's deaths have almost doubled from a week ago it seems like, which is really bad because the ratio of deaths-to-official-cases was really very high, indicating a vastly higher case count than is reported.

On the topic of vastly higher case counts than reported is Sweden. Sweden has 12,500 cases, but 1333 deaths. Their ratio of cases to deaths is actually much worse than Brazil's right now, and Brazil was though to have several times the official caseload by scientists there. (The ratio of cases to deaths in Brazil is 6.32%, the ratio in Sweden is 10.63%). That's... mind-blowingly bad actually. Implies either the Swedish Healthcare system is 40% worse than Brazil's (which I don't find likely), that the Swedes have some sort of super-strain which is straight up 250% as deadly as the regular virus (which I really really hope is not the case), or they're dramatically undercounting their totals.

You have to look at the #of tests in each country. Specially per capita. IIRC Sweden is low on that list. No superstrain, just superspread and infradetection. Which is happening everywhere tbh but the less you test the worse it looks and the worse you control it.

Mind you, while Chin and Russia are in the champion's league of releasing incomplete covid19 information, I think every country is doing fairly terrible at reporting the nunbers of infected, dead, tests, and the criteria for each. So take everything with a pinch of salt

[Starver]

I'd like an interactive visual graph, for each country, along the lines of the Y axis is %age of population, logarithmically; X time since given number of <foo>, which can be adjusted externally, like the currently popular base of "since 100th case", but 'case' could be test/death/etc.

It'd then be primarily a stacked data lines/areas being (baseline to top) something like mortalities not obviously associated with COVID, mortalities with unconfirmed suspicions of COVID, mortalities ascribed to COVID (would be nice to split that up into serious pre-existing comorbidity), recoveries, individuals tested positive (not part of the prior two at that datum), tested individuals not yet resolved at that datum, those confirmed negative (at that time, does not preclude their retesting later) and then up to the 100% ceiling would be all untested individuals.

A dotted line can also be sent up there equating to total number of tests conducted (retested individuals would raise that figure higher above the cumulative "tested" results level, deaths without tests would drag it down a bit and it could eventually get beyond the 100% ceiling if there's a concerted effort in a given population). That's for current infection, but another trend-line for historic infection/immunity testing could also feature importantly.

Mark times of major policy changes (lock-downs/open-ups, whole or partial, etc) with handy markers.

And then, with transparency of layers, allow comparisons between two (or more?) overlaid charts. Could be countries, regions, continents, towns, global figures or single naval ship. Though with the smaller populations being noticably more granular you'd expect steeper lines to wipe out the untested+live upper section, as a consequence so that's one thing you should not be surprised to see..


Not all exact measures will be available, and some may be lagged, but additional markup can show this. On top of not having any hooe of knowing anything for sure for daily updates, so the most recent edge would fade into blurred extrapolation, ) countries with revising counts (as China appears to have done) or changing metrics (UK implicitly counting 'hospital deaths' only, but we could easily post-facto shove in non-ward deaths under whatever class best fits) are already back-extrapolatable to indicate what is and is not a blip like a single immutable line would suggest.


It all works in my mind, anyway. It's not so much the absolute figures we're supposed to take from this but the relative profiles (and thicknesses) of zones. An ahistorical Non-Covid Deaths area and/or practically invisible (even on log-scale) thickness of all Tested sections would indicate likelihood of any problems with a certain jurisdiction. (c.f. the Floridian counties, as mentioned, or those countries recently mentioned).

[sluissa]

I've seen some people make attempts at charts like that, but usually end up quitting in frustration because the raw data is just so freaking bad. Testing numbers are shit so you can't get representative samples from ANYONE. Everyone reports different numbers and has different criteria for who fits into what category. And even when you do have something that should be easy to chart like a date of lockdown or reopening, it's kind of pointless without knowing exactly how each individual area was actually enforcing said lock down. Some places said things were under lockdown, but effectively weren't.

I think at this point we really can't say we understand truly how bad it is. And unless things change drastically, I don't think we'll know until it's all over.

[Starver]

it's kind of pointless without knowing exactly how each individual area was actually enforcing said lock down.

Reverse that. The chart (with the caveat you give of obvii6sly being problematic to create, which I agree with) might actually give you an indication of how much each lockdown is enforced that you never had before, through the relative and comparative inflections.

[Max™]

https://en.wikipedia.org/wiki/Geography_of_Russia#Area_and_boundaries

Area - comparative:
Slightly larger than twice size of Brazil

I think you meant more populous/densely populated, sorry, brain does that thing where it's like "ERROR: larger country is not half as large" when that happens. As for population, 211 million vs 146 million is probably what you were talking about, unless you were actually just thinking about butts, because those brazil asses be phat yo!

As for counts, I just check this regularly: https://www.worldometers.info/coronavirus/

[hector13]

aaaaay

2 months in and my employer has said all the employees need PPE (provided by them) to work.

aaaaay

[Starver]

Ask whether that would be specifically a one-eyed one-horned flying PPE, or any old kind...

[hector13]

You made a funny. Very good.

How long have you been waiting to roll that one out/how many times have you said it?

[Starver]

The lead-time was shorter than the time it then took to tap it in. First time (and rather unpolished, I think[1]).

What surprises me is that a) I hadn't thought of it before; b) I have not seen/heard anyone else do that. So obvious, in principle. Or maybe no budding wit thought it worthwhile.

(Nor did I hear any playing with the term applicable to virtually all top-line modern politicians back when Boris was rather ill. Maybe I just don't hang out in the wrong places that make potentially sick jokes about definitely sick patients.)



[1] The scansion needs a bit of work, but I think I could write it up as a filk lyrics. Not sing it though. Even the bit behind the anonanimity of the speed-changer. And, again, it's surely been done already.

[Jimmy]

aaaaay

2 months in and my employer has said all the employees need PPE (provided by them) to work.

aaaaay

You might want to check if this is legal in your local state/country.

For example, in Australia, if an employer requires an employee to wear PPE during their duties, they're obligated to provide the PPE or pay the cost of it.

I'd suggest a written request along the lines of "Dear management, due to shortages of PPE caused by coronavirus, my local supplier has advised they have no stock of these products available. In order to comply with your new employment requirements, please advise on your preferred location to source reliable supply of PPE. Alternatively, please supply these products directly to our location for staff use in order to meet these requirements."

Under our local laws, this would be enough to launch an unfair dismissal claim against an employer should they create a condition of employment but fail to adequately provide the means to meet that condition.

But again, this is relevant only to my local situation, so it's best you investigate your employee rights where you work.