Good things to read in this context:
“Being Mortal” by Atul Gawande.
“The Median is Not the Message” by Stephen Jay Gould.
BTW, from the moment Obama said he wanted all of American medicine on computer, we have dreaded the day when they would start data dredging and numbers crunching in order to generate yet more government guidelines.
Several considerations here:
1) Some people are on government Medicare, Medicaid, CHIP, VA...government programs. Others are in heavily regulated private insurance. Others are in medishare and direct pay which have much less government regulation. Others have nothing. They are self-pay til they run out of money.
Should end-of-life be different depending on where the money comes from, who pays? A pure libertarian says nobody should be forced to pay the medical care of another person. We will never get that libertarian. But what about a modified libertarian position? We will force taxpayers to pay for care up to a certain threshold and no more. After that, if private charities, family members, etc want to pay they can. One could buy an end-of-life insurance rider to a normal insurance policy.
2) Garbage-in-Garbage-out. I deal with healthcare data, data warehouses and statistics as my career. There is a lot of bad data. Even with good data, famous healthcare analytics firms that can play chess and Jeopardy can’t do 8th grade math in healthcare. What level of error would we allow in predictive analytics for life and death decisions? 1 in a million? 1 in a thousand? 1 in ten? 1 in 3?
3) Predictive Analytics, especially in healthcare, is filled with a lot of subjective bias in the models. Is the life of an important politician or business man or much loved celebrity more valuable than the life of a non-productive person who has been on welfare all life? or more valuable than a person who has enjoyed white male privilege all his life and now his privilege should run out? Is the life of a prison inmate less valuable? the life of an illegal immigrant? A refugee? Who decides the criteria?
4) The issue is much broader than just end-of-life. The Medpage website recently published an article on medical decisions in prison. The doctor argued that most prisoners have liver disease. But most who have liver disease will not be in a critical stage for many years in the future. They should be ignored and only those with severe conditions who also have the prospect of successful treatment should be treated.
Because treatments take many months, prisoners in for just a few months should not receive treatment because they will be released before the treatment can be completed. Of course, taxpayers pay for the treatment of prisoners.
5) Predictive Analytics are taking over many fields. It is a tool that can be used skillfully or unskillfully. And it is just a tool. Will Predictive Analytics replace “I saw it on the internet. So it must be true.” or “I learned it in Public School. So it must be true.”
Predictive Analytics says Clinton has a 95% probability of winning and Global Warming has a 100% probability for a 1 degree celsius increase, 80% for 2 degrees, 50% for 3 degrees. Predictive Analytics is based on models. AI is just Predictive Analytics on steroids... and still based on models. Those models are built by humans with bias, as we who follow politics are well aware.
Will we let Predictive Analytics take over our society? Why have an expensive election if Predictive Analytics can acurately predict the winner. Nate Silver just needs to tweak his model a little bit for 2020 and we could avoid a lot of expense and hassle.
*The percentages for Global Warming are illustrative only and not actual.