WANTED: Rational Coronavirus Analysis

#1
Continuation of thread: http://www.talkstats.com/threads/simultaneous-convoluted-linear-equations.73663/#post-218486

The overuse of percentages by our wise ruling class and media to convey critical information may be hindering understanding, and therefore a rational response to the Coronavirus pandemic. Lots of figures are being thrown about, practically all on a percentage basis, not as ratios, as natural phenomenon should be expressed.

This STAT News article (March 3rd) falls into the same base-10(0) percentage rut...


...but at least they're slicing and reporting the data well, which is quite rare these days.

Excerpt:
The death toll skews old even more strongly. Overall, China CDC found, 2.3% of confirmed cases died. But the fatality rate was 14.8% in people 80 or older, likely reflecting the presence of other diseases, a weaker immune system, or simply worse overall health. By contrast, the fatality rate was 1.3% in 50-somethings, 0.4% in 40-somethings, and 0.2% in people 10 to 39.

NOTE: The Chinese CDC data may not be totally credible, but that uncertainty factor would apply to all categories equally, presumably(?).

The ratios on total-infected per fatality (the inverse of the above percentages) for COVID-19 by age follow:

80 and older ..... 7
50-something ... 77
40-something ... 250
10-39 ..................... 500

Analyze this ... again, with ratios. It's lucky that we live in a country with such robust youthful leadership, otherwise policy might be skewed.

And, lastly ... Don't shoot the messenger ... you've been advised to self-quarantine, and will be reported to authorities.

In related news: https://babylonbee.com/news/nations-nerds-wake-up-in-utopia-where-everyone-stays-inside-sports-canceled-social-interaction-forbidden
 
Last edited:
#3
I must say, that reasoning is hard to argue with. I can't even re-phrase it as a formula. The best I can come up with...

Numerous (Twitter infographics) = "Nice"​

A Venn diagram is probably the best way to express your position.
 
Last edited:
#4
This Stanford study on COVID-19 is very comprehensive, informative and level-headed: https://drive.google.com/file/d/1DqfSnlaW6N3GBc5YKyBOCGPfdqOsqk1G/view

While not posted on Twitter, it has some nice infographics, including fatality rate by age for South Korea, which sees the same basic demographic impact as China, heavily skewed towards the elderly:

Korea--COVID-Death.rate,0-3.jpg
The South Korean ratios on total-infected per fatality (the inverse of the above percentages) for COVID-19 by age follow:

80+ ....... 12
70-79 ... 21
60-69 ... 69
50-59 ... 250
40-49 ... 1111
30-39 ... 883
<30 …... Infinity

So, no surprises* ... which I guess is good, in uncertain times. In addition, experts are estimating that perhaps half or more of the infected are asymptomatic, and don't bother to get tested, so this would increase these infected-per-death ratios even further.

* Except for vulnerable 30-somethings in Korea..?
 
Last edited:

noetsi

Fortran must die
#5
Anyone could be right (or wrong) about this virus. No one knows for sure what it will or will not do so everything is speculative. It is difficult to analyze data rationally when almost nothing is known.
 

hlsmith

Less is more. Stay pure. Stay poor.
#6
If you are inverting, I wonder if it follow's Zipf Law. People feel Korea's number are the most representative if the under lying data generating function, since they tested a **** ton of people and even had drive-thru testing. China on the other hand is always dubious and they have authoritative regime, so can better contain. Lots of call info graphics for sure. I will start a thread on it. I pulled the Hopkins data down, which seemed to have a lag, but will work for playing around with data.
 
#7
"Anyone could be right (or wrong) ... No one knows for sure ... almost nothing is known."
Gotta disagree with you there, noetsi, and think that hlsmith would agree: Independent studies are showing that critical trends are known. 'Not acknowledged' is probably the proper descriptor. And if this is a case of 'willful unknowledgment', that would be especially unhelpful.

The Korean tally is the better set of data, but does conform with what was experienced in China. Yes, some statistical analyses may be applied, yielding some insightful infographics, perhaps independently threaded on Twitter.*

Meanwhile, on the other side of the world:

Italy has had 12 462 confirmed cases according to the Istituto Superiore di Sanità as of March 11, and 827 deaths. Only China has recorded more deaths due to this COVID-19 outbreak. The mean age of those who died in Italy was 81 years and more than two-thirds of these patients had diabetes, cardiovascular diseases, or cancer, or were former smokers.

I'd say that 'median age' might have been the better metric here, but suggest that this data agrees with what was seen in the Far East, with the added critical dimension on susceptibility with pre-existing conditions.

In the government-mandated world of not-known-so-anything-goes, where incubated-carrier-but-not-at-risk college students are sent home for the semester, since Reasons, et al ... I'd suggest that those seemingly-healthy bored kids don't hug their 85yo great-grandma, the diabetic chain-smoker ... even if she is Italian. Can we agree on that?

* But not by me. My stats/math training didn't really extend past my teenage years, and I'm still not sure how Twitter works and why.
 
Last edited:

Karabiner

TS Contributor
#8
They are vastly imprecise. Only 2 decimal places?! Given that they absolutely, exactely know the
number of infected persons and absolutely, exactely know the number of deceased persons in
Korea, their death rate should comprise 4 decimal places. At least. Rounding looks very unscientific.
People could be tempted to think that science mght deal with uncertainty.

SCNR

Karabiner
 

Miner

TS Contributor
#9
I would disagree that they exactly know the number of infected people. That would assume that they have tested 100% of the population and that the test had zero false positives and zero false negatives. Good article by Kaiser Fung that illustrates all of the unknowns.
 
#12
Yes. It was meant ironically.
Irony doesn't work well in a forum, though.
Yes, I didn't catch the irony either ... 'twas ready to note the existing vast uncertainty in Covid-19 prevalence within the untested general population. The number of infected people that were tested is known; what is not known is the number of infected people not tested.

Anyway, good one. Reporting data to statistically-meaningless significant digits should always be considered fodder for humor ... well, at least 99.99% of the time.
....
Meanwhile, STAT News reruns their age-based analysis, again with appropriate significant digits:


--------[ WAIT ... THERE'S MORE!!! ]-------

The Hoover Institute is taking measures to de-tangle the data, focusing on age (including a detailed breakdown on the Korean numbers, with a bit too many significant digits on fatality rate) : https://www.hoover.org/research/coronavirus-isnt-pandemic

Korean.fatalities,0-3.jpg

Now, our America-on-lockdown-quarantine CV efforts are starting to re-scare China:
Hey, we didn't see that coming!

Fatality rates by age are pretty consistent in China, Korea and the US (charts towards the end):
It's all about pre-existing conditions:

Wish this icky guy would stop making sense ... makes one forget what the CDC and WHO and Bill Gates told us to do...

Wishful thinking: https://www.takimag.com/article/how-do-we-flatten-the-curve-on-panic/

The massive global CDC/WHO effort StatsNews continues to be THE forefront of understanding the scope of the Coronavirus pandemic:
Understanding data statistically is good, and, yes, news based on statistics is critical, providing clarity ... helps one understand inherent probabilities, even without associated odds. (Ironically, compared to their earlier posts, this StatNews article is short on statistical analysis, but that never stopped media outlets.)

This is encouraging, New Media is considering the 'Big Picture' ... Ezra Klein at Vox is starting to ponder rational CV plans that consider the economic damage (check out the unemployment chart, yikes!):

Also encouraging ... (very non-mainstream) commentary with rational Coronavirus analysis:

Israel also has consistent fatality data, by age:

At the very least, this could lead to solid ratio-ed analyses:
Excerpt: They concluded that roughly 80% of the people who lost their sense of smell would test positive for the coronavirus, and that somewhere between 30% and 60% of those who had tested positive for the virus had also lost their sense of smell.

Dubya foresight, revisited:
Excerpt: In a November 2005 speech at the National Institutes of Health, Bush laid out proposals in granular detail -- describing with stunning prescience how a pandemic in the United States would unfold. Among those in the audience was Dr. Anthony Fauci, the leader of the current crisis response, who was then and still is now the director of the National Institute of Allergy and Infectious Diseases.
Too bad the message was lost for years, now dredged from the archives.

More perspective, from a 'Big Picture' guy with street cred on megatrends:
Excerpt: “Universal stay-at-home is the most devastating economic force in modern history,” Burry wrote in an email to Bloomberg News. “And it is man-made. It very suddenly reverses the gains of underprivileged groups, kills and creates drug addicts, beats and terrorizes women and children in violent now-jobless households, and more. It bleeds deep anguish and suicide.”
Fatality data by region, age and pre-existing conditions:
NYC is an outlier in the Covid-19 pandemic in that fatality rates in the not-so-old population are higher than elsewhere: about 30% are under-65 and about 46% are under-80. That said, in the NYC U80 fatalities, less than 2% of deaths did not have pre-existing conditions.

The disparate impact of globalization: https://www.takimag.com/article/ill-have-the-chicken-testicle-soup-hold-the-deadly-virus/
Excerpt:
[COVID-19] deaths, so far, by population:
— New York (9% Asian): 29 per 100,000
— New Jersey (10% Asian): 13 per 100,000
— Montana (0.9% Asian): 0.6 per 100,000
— West Virginia (0.8% Asian): 0.2 per 100,000
Don't doctors encourage 'second opinions'..?
Excerpt: You … have functionaries – all of whom are still getting paychecks – who are way too eager to start issuing orders instead of relying on citizens to behave rationally.
…what the hell they are thinking telling us what is and is not “essential” anyway? They are thinking, “Hey, this power stuff is fun!” Which is why you should never give power to someone who enjoys it.
But the medical perspective is not the only perspective. There are other perspectives that need to be considered, like the economy and our liberty, and so we elect people to consider them. See, this is why we are not a technocracy. We are a republic. We normal citizens get the final say in our country’s priorities, and we may not decide that our priorities are exactly the same as Dr. Fauci’s. I, for one, am willing to accept some risk regarding the infection in order to avoid America degenerating…
Uncertainty and risk requires rationality to properly assess, and to formulate the pathforward beyond STOP EVERYTHING UNTIL FURTHER NOTICE.
 
Last edited:
#14
News items are now popping up, squabbling about how Coronavirus does infect the young, so there, we’re all in this together. Of course, the relevant metric remains steady … fatality rate by age shows who’s at greatest risk: the old and/or ailing.

So where to focus our attention and resources, when so little is known about what’s actually known?

Perhaps we can go beyond a simple ‘futures ratio’ -- Young/Old, a rational number -- and apply a nonlinear zero-sum algorithm to roughly capture the monatomic dynamic, which can be viewed as equivalent to the odds in a two-outcome competition:

(Young folks’ futures) x (Old folks’ futures) = 1​

If this is the case, investing in and focusing on increasing Old folks’ futures (which are naturally more limited, due to lifespans and ailments) by 10% decreases Young folks’ futures by about 9% (1.10 x 0.91 = 1).

How might that be manifested? Well, aside from the direct costs of medical care, how about … the cost of shutting down the economy to ‘control the virus’ will necessitate massive amounts of government spending to mitigate the damage to workers and businesses, which will result in large debts that need to be paid in the future … by the working young, of course, not the retired old. And that’s just one manifested destiny.

In addition, managing this wealth-transfer process has its own inherent costs and burdens, which can be summarized as t, for the house take (and this ‘house’ makes and enforces all the rules and prints the money).

(Young folks’ futures) x (Old folks’ futures) = 1 – t
NOTE: The factor t includes administration, compliance and debt-burden costs, all of which are already big and getting much much bigger.
Finally, in the interest of Don’t shoot the messenger/theorist … if you think that this elegant proposition is not at all helpful, and merely a ‘hate algorithm’, that may say more about you than anything else. Which, in itself, could be an algorithm ... or fractal, maybe. [Insert appropriate emoji]

For a survey of sundry monatomic relationships, see: http://www.talkstats.com/threads/nfl-postseason-probabilities.74650/#post-219049
For background on the monatomic relationship of competing zero-sum odds, where Odds.1 x Odds.2 =1, see (after video in post): http://www.talkstats.com/threads/nonlinear-odds-to-probs-conversion.73716/#post-217159
 
Last edited:
#15
A paper on "Tail Risk of Contagious Diseases" by Cirillo and Taleb (link below) has recently been released, and notes that "pandemics are extremely fat-tailed, with a potential existential risk for humanity". The authors state that to the best of their knowledge, "only war casualties and operational risk losses for banks show a comparable behavior [to pandemics], and they are both phenomenon very difficult to model."


This effort by the authors and their assistants goes way beyond conjecture about analysis methodology and projection of possible scenarios. They have consolidated the hard numbers from pandemics throughout history into a very comprehensive list. These are events that have already happened, and the paper puts them into relative perspective by normalizing fatality rate with today's worldwide population (Table 1, Rescale Avg Est).

The table below was derived from their Table 1 data. It gives the specific pandemic, its active dates, the actual fatality rate (average, estimated), and a 'rescaled' fatality rate that normalizes the pandemics' death rates for the same relative impact on today's worldwide population. The order-of-magnitude of the various pandemics in comparison to today's Coronavirus crisis is shown. As natural-world functionality is generally nonlinear, that metric used here (but not in Cirillo) is the number of times the fatality rate of the COVID-19 death rate would need to DOUBLE to match that pandemic's death rate within the overall worldwide population. The data is sorted by this fatality-per-capita order-of-magnitude. For instance, at the extreme, the Black Death had a worldwide impact that was about 54,000X (or 2^15.7) greater than today's COVID-19 pandemic to date.

1586186525047.png
Of course, the ongoing COVID-19 pandemic will only climb this chart, as more and more fatalities are recorded, probably having a 'fat-tail finish', as the authors note. But this comprehensive study does offer cold-eyed perspective. Right now, COVID-19 has already outpaced the localized and exotic outbreaks (most fairly recent), and is now solidly in the 'Malta plague' range in relative severity. (Malta has been an international hub for millennia ... Sir Stamford Raffles, the founder of Singapore, had hoped that the swampy island territory would become the "Malta of the East." This brings up an important relevant factor: Pandemics often start in 'trading posts', where folks from vastly different regions meet, hence the risk of spreading deadly disease.)

In reference to the above 'cost/benefit Young/Old' theory (previous post within this thread), an imaginary column is shown on the overall worldwide 'Societal Impact' that each pandemic had on humanity. What academic discipline can be brought to bear in establishing and estimating relative metrics on Societal Impact of pandemics, historically and today? Perhaps there is a indirect metric, that is simple, historically durable and consistent, and is recorded faithfully by a steady societal strata. Tax paid to the state, generally inescapable and duly recorded, might be a good one. When times are good, tax revenue is high ... when times get tough, tax revenue suffers (along with the people).

........

It is imperative that we learn and gain perspective from history. Perhaps there are some lessons from the Malta plague of 1813 that can be applied to today's crisis:
 
Last edited: