Can Artificial Intelligence Cure Healthcare?
I am not an expert on health care nor am I an expert on artificial intelligence, so I am just asking some questions here. But it seems to me that our American healthcare system is a complicated mess, and it is a complicated mess because of people. People make things complicated because different people want different things. And some people (and some institutions) have greater influence that allows for special cases, exceptions, carve outs and subsidies that make something nominally fairly straightforward into a complicated mess. Like our tax code, for example. That’s also the reason the recent omnibus appropriation bill was 4,155 pages long. The implementing regulations will be many more thousands of pages spread across all the agencies of government. Each senator and representative voting for the bill got a little something (some got a lot).
Health care is a lot like the omnibus appropriations bill except that it has had fifty plus years of government interference to concoct our current mess. US healthcare employs millions of people and consumes almost a fifth of our economy. As noted in a recent commentary (Non-Trivial Pursuits, January 23, 2023), the Congressional Budget office projects that healthcare will make up almost a third of our economy by 2035, converting the country into a gigantic hypochondriac incapable of self-defense let alone global leadership.
More government is not the answer, as was shown in that same commentary. European government run healthcare, once highly respected, is now falling apart. This is not surprising. Governments are good at some things and not so good at other things (in fact, they can be downright awful). A USA Today study, for example, found that 141 out of 146 VA hospitals scored lower than other nearby hospitals. Government doctors and nurses metastasize into bureaucrats over time. Paperwork becomes more important than patient outcomes.
Of course, totally unregulated medical care isn’t the answer either. Such healthcare would be superb for the wealthy and elites, but mediocre or poor for everyone else. The problem with healthcare is that it is complex. Doctors have to study for many years to become proficient in their field (and there are many fields of specialization). Healthcare consumers are not well educated on these matters (how could they be?), so they lack the ability to make wise healthcare decisions.
It seems to me that artificial intelligence (AI) is ideally capable of cutting through all this complexity. AI thrives on complexity and big data. ChatGPT can write your doctoral thesis in some obscure study of physics or neurobiology. DALL-E 2 uses its library of hundreds of millions of images to create any sort of new image you can imagine. DALL-E 2 created the image used in this commentary. If anything can cut through the enormous complexity of healthcare it must be AI. Already AI is being used to identify cancer cells imperceptible to most medical specialists.
By cutting conniving politicians, greedy corporations, woke professional associations, special interests and advocacy groups of all sorts out of the process, Artificial intelligence would have a much better chance to create a healthcare policy that actually works. All we have to do is describe the healthcare system that we want, and the AI will do all the rest. As if it were really that simple.
Well what kind of healthcare system do we want? We have to tell the artificial intelligence what end result we desire so that AI can create the policies that will help us achieve those goals.
Firstly, and most importantly, we want a system that will give us good patient outcomes. Not unachievably ideal outcomes, but good achievable outcomes. We want a system that is innovative, because it is only through innovation that we can apply our growing knowledge of medicine so that we can improve patient outcomes. We need a system that is efficient. The healthcare system cannot absorb too much of our resources because we have other goals in life that we also need to achieve. We need a system that encourages healthy living and does not reward those who choose unhealthy habits. We need a system that if fair to the poor but does not reduce the healthcare system to the lowest common denominator. We need a system that does not sacrifice science to ideology but informs ideology through scientific inquiry. Most importantly, the new healthcare system must incorporate safeguards to protect it from politicians, special interests, and other powerful institutions from creating carve outs, subsidies, exemptions and other self-serving anomalies that will undermine the system and would ultimately destroy it over time as the anomalies accumulate and overwhelm the operation of the system. That is the situation that we are currently in and that we wish to leave behind us.
Can we achieve such a wonderous healthcare system? In the past, we used to say, “God only knows.” But maybe now we can say, “Only Artificial Intelligence knows.”
But how we instruct AI to create a new healthcare system will depend on how we frame the instructions to the computers. And because humans are not infinitely adaptable, we will need AI to also create a transition mechanism to move actual people from the old system to the new. And further, as Professor Daniel Kahneman and other psychologists have repeatedly pointed out, human beings are not entirely rational so the system must incorporate appropriate incentives that will encourage people to help the system achieve the positive patient outcomes we seek.
It seems to me that a successful healthcare system would require a number of algorithmic parameters to enhance the chances of success. It would have to be based on the free market economic system because only free markets provide the innovation and efficiency needed to constantly improve healthcare while lowering costs. Heath consumers (that’s all of us) need to have some skin in the game. Only if they have to pay a portion of their medical bills will consumers be forced to consider both price and outcomes. Patients would have to be more responsible in selecting treatments and healthcare providers, but AI can use its enormous database to help with this process. Medical insurance should actually be insurance, that is pay for emergencies and unforeseen events and not routine doctor visits and minor infirmities. And the insurance should belong to the individual and their family and not be a company provided benefit so that the individual can change jobs without losing medical coverage. If the company wants to help pay for medical insurance in order to entice prospective workers, that’s fine, but any tax deduction the company gets should be offset by counting the benefit as taxable to the individual (the double tax exemption of company provided healthcare is one of the principal factors that created the current distorted healthcare system).
There are other things that we must do. The high cost of prescription drugs in the US is principally due to other countries free riding on costly American based R&D and rigorous testing that companies can only recover in the US Other countries only cover production costs by threatening to pirate US patented medicines. And some people will be unable to afford even affordable healthcare so some form of assistance should be made available as long as it doesn’t disrupt the entire system. We would need to discover a way that people living healthy lifestyles don’t have to subsidize those that don’t (such as non-smokers paying for smokers lung cancer treatments). And we need to figure out how to pay for genetic disorders and pre-existing conditions (as well as ways to determine what constitutes a pre-existing condition or genetic disorder).
A lot of what is currently being treated as a medical condition is actually ideology and medical schools and professional associations need to purge ideology from their organizations and get back to science. Maybe artificial intelligence can also help separate racially disparate disorders that are genetically based (such as sickle cell anemia), from lifestyle choices, cultural differences and actual cases of discriminatory treatment.
That is a lot to ask of artificial intelligence, but I would rather trust AI than conniving politicians, race-baiting community organizers and leftist progressives and academics.
Comments