Healthcare AI Does Not Need A New Hippocratic Oath
This article explores the historical context of ethics in healthcare technology, highlighting the relevance of an ancient oath and the need for a strong ethical framework in AI.
Ethics in AI is a major issue, especially in response to the role that artificial technology has played in growing hate crime, destabilizing democracy, radicalizing youth and scaling up discrimination -- all while profiteering from it.
This is of utmost concern when life-and-death decisions are automated -- as in healthcare. This resulted in broad discussions about the need for data scientists to take a form of the Hippocratic oath -- or even that AI solutions should take it in some form.
However, if you do work in healthcare, you'll know that ethics in healthcare technology is much further along than in software technology. I challenge you to live by an older oath, which also has the benefit of putting the AI revolution in a much humbler historical context.
Healthcare Technology: Tale as Old as Time
Pharmacy -- practiced since the beginning of civilization and documented since the beginning of writing -- is the first form of healthcare technology. Instead of directly seeing patients, a pharmacist would concoct potions, pills, creams, and devices that patients or physicians would buy. They claimed to relieve pain, get you back on your feet or make you younger, prettier, happier, or sexier.
The top sellers never change.
These technologists would concoct whatever remedy they could at home and then bring it to market. There was usually no proof that these products worked at all -- and every financial incentive to lie, exaggerate and distract people from facts. Great marketing -- showmanship, packaging, naming, distribution -- was a blueprint for commercial success throughout history.
It didn’t help that until the development and wide adoption of statistics, there was no way to test what really worked. New technology -- whether it was a new medicinal herb or surgical technique -- was always unproven for a long while and always attracted many more charlatans than true believers. We’re at that point today with digital health.
These ancestral healthcare technologists did not explain how their products worked because that secret was their livelihood. Often they didn’t know themselves. Explainability is not an issue that started with deep learning -- it started before the alphabet.
There was never a time in history where ethics in healthcare technology was not an ongoing public concern. Bad actors kept selling, and good people kept buying -- because of the placebo effect, because there’s a new sucker born every minute or because the sick just wanted a glimmer of hope.
Pharmacy And The Oath Of Maimonides
Pharmacists -- those in charge of the safe application of healthcare technology, from medicinal herbs in Sumer 5000 years ago to prescribing the right app to the right patient today -- already have a strong ethical framework in place.
The most widely used oath is the Oath of Maimonides. Attributed to the 12th-century rabbi, physician, and philosopher of its namesake and published in Germany in 1783, it is surprisingly appropriate for healthcare AI practitioners today. For those unfamiliar with the oath, here is a snippet from the full text:
"May the love for my art actuate me at all time; may neither avarice nor miserliness nor thirst for glory or for a great reputation engage my mind; for the enemies of truth and philanthropy could easily deceive me and make me forgetful of my lofty aim of doing good to thy children.
"May I never see in the patient anything but a fellow creature in pain?"
"Grant me the strength, time, and opportunity always to correct what I have acquired, always to extend its domain; for knowledge is immense and the spirit of man can extend indefinitely to enrich itself daily with new requirements."
"Today he can discover his errors of yesterday and tomorrow he can obtain a new light on what he thinks himself sure of today."
The Way Things Are
As an industry, we are far away from living by such an oath. Consider the distance between never seeing anything but a fellow creature in pain and these statements:
In a fee-for-service setting, a patient walking into the hospital is a revenue stream.
In an accountable care (ACO) setting, the same patient walking in is a cost center.
Payers are the true customers of a healthcare provider -- they are the ones who pay.
The terms "patient" and "healthcare consumer" can be used interchangeably.
The quadruple aim of healthcare balances trade-offs between improving the patient experience, improving population health, reducing the cost of care, and improving clinical staff satisfaction.
This is, unfortunately, how U.S. healthcare speaks in the early 21st century. It’s so pervasive that it was impossible for me to pick citation links here -- just search for top-of-mind statements from executives and leaders of provider groups, payers, startups, health IT companies, pharma companies, and everyone in between. It’s a business, first and foremost. Love of the art is second to making money or fame, and for every Theranos that gets caught, many more keep going.
There is nothing new under the sun.
The good news is that as a data scientist, you do not need to wait for an ethical framework or for widely accepted solutions to AI bias, explainability, or plain misuse. Simply build to help patients and humbly acknowledge the limits of what you build. Keep learning. Join a team that does not require you to make unethical trade-offs.
Not there yet? Today you can discover the errors of yesterday, and tomorrow you can obtain a new light.