Data science and COVID-19

Tracking by mobile providers do not bother me so much. Data sold is supposedly anonymized.

What does concern me are things like this:

And a constant drip, drip, drip of stories like this:

1 Like

I believe Eric Snowden proved that the NSA is the best suited to provide this valuable service world wide. They already have the data and the computing power. Uh oh, I hear the black helicopters now!

2 Likes

A bit morbid, but I have been watching this page for the USA deaths counts. Curious to see if it actually follows the projected trend or if it drops off sooner.

https://www.worldometers.info/coronavirus/country/us/

1 Like

Depends on whose projection.

The curve of confirmed cases is nowhere near bending for the US.

Washington state seems to have it under control.

I guess the one I was talking about was that the expected peak is April 15th at around 2,500 per day. I think it would be interesting to see how certain states compare since some seem to be doing more than others. Curious if actually ends up being about that date or if it extends past. I would say I’ve never seen anything like this, but I think there are some folks older than me that could say the same thing.

Two weeks is an eternity. Some “experts” expect Dallas to be a hotspot in that time.

I have not been out, but hear horror stories about people not socially distancing at grocery stores.

I am an old fart, but have not seen anything like this either.

Right after 9/11, we waited and there was no second wave of attacks. We went back to work.

The financial panic of 2008 was scary, but we went back to work.

This time, many will have no work to go back to.

2 Likes

Ive been doing the same thing for some time now. New York is not far from passing Italy & Spain for cases.

When you get older you’ll check the obituary pages to see if your name is there … then smile when you don’t find it.

1 Like

As I got older, I used to regularly check the obits of the hometown paper to see if a classmate or their parents pass.

I once discovered someone who used to work at our restaurant. I told Mom, she had not heard.

After Mom passed, I quit checking. I had become one of them.

I find it more meaningful to watch the rate of change in the numbers rather than the raw counts. This interactive log-log graph is particularly enlightening.

2 Likes

This is helpful, I hadn’t seen this one yet. (Besides something from that YouTube video)

3 Likes

This was the graph mentioned in the video shared here recently. I like how you can drill down to selected countries.

1 Like

My wish for the follow up to this project with Minute Physics and Aatish Bhatia is for it be updated (frequently, I wish for, daily at least). I lack the expertise.
But they did a really nice job making the data more meaningful, more interesting, and explaining why and how it is more meaningful/interesting (as well as covering some of the drawbacks and explaining why every datagraph is flawed).

EDIT: actually, I just realized this HAS been updated since the video was released, with NY being up to 3/31/2020… Maybe my wish has been semi-granted (daily would be nice, but I know people have things to do…)

I don’t think it is morbid at all but rather scientifically more sound than tracking reported positive test results.

Tracking reported cases is not all that useful because we are not randomly sampling. Rather we are testing those with a higher likelihood to actually have the virus; healthcare workers, those with severe symptoms, etc. In addition the number of tests administered per day has been exponentially increasing. All it does is help the politicians scare the population into compliance. While that might be a good reason to report test results, it does little to help with actually measuring effectiveness of policies like social distancing.

On the other hand tracking deaths is more useful because there is near 100% reporting. I really wish we were tracking hospitalizations. Reported hospitalizations should also be near 100% because they are nearly all confirmed with tests. We can track either deaths or hospitalizations and look for a tail off in number of new reports / day. But tracking reported cases, given the flawed sampling, is only useful to those that want to show a sharply increasing graph

1 Like
1 Like

Wouldn’t it be so much more fun to find it there? Like these people did:

Interestingly enough, my uncle was prematurely declared dead by the social security office and had to prove to them that he was still alive. Still don’t know how that mix-up happened.

1 Like

Testing is not perfect. The cdc recommends absence of fever for 72 hours and TWO negative tests 24 hours apart for discharge of patients with confirmed disease. That should give a clue as to how they feel about the specificity of the test. I know of one patient who had all the symptoms of covid including anosmia that tested negative. I agree mortality is a more reliable indicator.

1 Like

Interesting article that also has links to other background info:

1 Like

Three friends came back from overseas when this all came down. They filled out paperwork…pen and paper for immigration in nyc Florida and dfw. Why wasn’t it all digitized?

Also, I saw something of a qr tracking system implemented in Asia. Everything…buses,cars,buildings have a unique QR code. When each person enters they scan it…that info can quickly be cross referenced into a database for tracking.
Patient x tested positive as was at the store at the same time/within 2 hours of q,r & s…send out a message to all of them.

Yes yes yes I hear you “my information is my privacy”

Just something to think on.

This is annoying, curious how far off we are

3 Likes