#POWEROFDATASCIENE - DON'T JUST TEST THE SICK!!!

Oh no, not another COVID19 post!! But, then this is different, hopefully you realize as you get through the write up.

Like everyone else, this weekend, my household also hit the pause button on life and the TV remote has not flinched off of news about the Corona Virus, even for a moment. Of course, I was multi-tasking as usual, and read my friend Sri Jagannathan’s LinkedIn post, while also watching Bill Gates’ 2015 TedTalk on global pandemic preparedness. And then it struck me that, that our Statistics 101 professor talked about way back when, and I am sure data scientists of the best caliber working in close concert with the CDC today, know – is being fully ignored, at least from the Covid19 Task Force and the President’s pronouncements on TV on Sunday!

Okay, so 1.5 million test kits are ready to hit the ground with over 2000 labs, public and private across the US testing the nasal swabs collected from the population tested. Now, this announcement itself is great, a very welcome relief to the bungled responses over the last 8 weeks, since the crisis hit. However, this announcement was followed immediately by another  fatal announcement, yes an announcement of fatal proportions, a statistical blunder that anyone with a basic understanding of statistics will call a big one!

The President announced that testing is restricted to those who are showing symptoms of the disease! That surely will achieve one result, treat those that are symptomatic so that we get a handle on the disease for those infected, treat them ahead of time, while the symptoms are mild and don’t let them get into the acute phase of the disease and control fatalities. YES! 

However, there goes statistical/data science principle #1 - getting an unbiased sample, else all analysis goes to the proverbial toilet! Testing the sick will enable us to treat them, or see the trees but will not let us see the woods. Any analysis based on such testing will exacerbate the results – X tested, Y infected from those, i.e. Y/X% is the spread of the disease??? All I have read, watched over the last few weeks is indicating that those are the only numbers being looked at, at least by the powers that make these decisions. What I see as of this morning on the most well designed ARCGIS representation is the following:

Total infected (globally) : 174,995

Deaths : 6706

Recoveries : 77,658

If we apply this method of decision making based on clearly wrong analysis of the underlying spread of the disease, we are causing huge PANIC in the markets and in society in general. Consider if we were to do the same thing with any other disease. Stretching logic to make the point even more clearly, say I test only those people that report to the hospital complaining of chest pain in my city, Pleasanton CA and that number is 100 in a month, of which 30 get admitted right away because they have suspected coronary disease and the others just had flatulence and were sent back home with an antacid. The 30 are put on treatment right away and 20 of them are sent home after a couple of days, while 10 remain in hospital of which eventually, unfortunately, say 3 die. Concluding that the incidence of heart disease in the population is 30 and fatalities are 3 which is 10% is such a mockery of all we know of science and statistics!!  

How many countries (South Korea?) have sampled from a statistically significant sample from the general population to extrapolate and assess the spread of the disease geo location wise, population age, density, socio economic status, etc. Even social distancing parameters could be included in such a study, and it could be run immediately in a matter of days, with the testing kits now (hopefully) available this week!

I am not saying don’t treat those that come down with the virus but launch an immediate statistical sampling-based testing in all areas, far and wide across the US with a statistically significant sample of people tested for the spread of the disease! Perhaps set aside 20% of the kits for such testing to get a sense of “where the puck may go next, rather than where it has been” to use Dr. Tony Fauci’s famous words from a few briefings ago. 

• That is how we will know what the real parameters of the spread of this pandemic is and how to deal with it.
• That is how we will ensure that stock markets, corporations, cities and states, don't take panic decisions founded on fear but based on unbiased, statistically and scientifically relevant studies.
• That is how we, the people will know how to lead our daily lives in these trying times, not hoard toilet paper and sanitizers and not hit the pause button on everything.

I am not a scientist, data or immunology trained, but am seeing decision makers being swayed in their decision making by some apparent wrong conclusions from data, hence this post!