Can "Expert Systems" (Medical Diagnosis) Solve The Healthcare Crisis?

The US Government is projecting a big shortage of primary care physicians.It seems to me, that sophisticated software for diagnosing medical problems might be an answer-years back, a computer program was written to simulate decision making in a very obscure skill in civil engineering-evaluation of dams. What I remember was that they used the expertise of the top dam engineers in the country, and then tested the program. The results were that the program correctly predicted the state of the dam, in a higher percentage of cases, that the experts themselves.
So could a similar program be developed for primary medicine? Most of my doctor’s time seems to be spend analyzing test results-a good AI program should be able to handle sch a task.

I don’t think we’ll solve this problem by eliminating doctors, no.

What we should eliminate first is third parties who suck money out of the system. They’re unnecessary and many other countries have done away with them, resulting in the same or better results and substantially less cost. Insurance companies are unnecessary, doctors are essential.

“Expert systems” do have a place, and I’d like to see them utilized more by doctors not just on the primary care level but on all levels. However, people are still better at dealing with anomalies and the unexpected. So… not ES to replace doctors but rather to make automated some of the boring routine stuff to allow humans to focus on the more interesting and unusual.

We could also utilize more folks like nurse-practitioners and physician’s assistants for routine things ranging from well person check-ups/screenings to minor infections and routine problems, flagging those that do, in fact, require a doctor’s attention.

Also: we could increase the number of GP’s by doing things like partial or full loan forgiveness and providing incentives for people to go into that field rather than a higher-paying specialty.

It’s purely a matter of corporate greed on the part of insurance companies and big pharma, it’s gonna take political action at the street level to break them. No software will do that.

“Expert Systems” - this was a specific approach to creating a domain expert in software that relied on a boatload of IF THEN type logic plus some other tricks.

I personally think that is a failed approach because it’s hard to spot patterns that way.

More recent software for detecting cancer etc. typically uses either neural networks or support vector machines which are tools that excel at categorization/pattern matching. They have had great success with these.

But it’s really just another tool to improve detection etc. - it doesn’t address why insurance rates keep going up.

At some point in the medium far future (within 500 years) both physical treatment and diagnostics will be done entirely by machine. Humans will be considered too dangerously imprecise to do surgeries and simply not intelligent enough to do diagnoses.

Not in any of our lifetimes, though.

Note that the healthcare crisis is really two things:

  1. Lack of access to care.
  2. Rapidly expanding costs for care.

Universal health coverage can control some of the costs and some of the access problems. The reason I say it cannot control all of the costs is even universal systems have healthcare costs rapidly increasing faster than inflation. The reasons for that are complicated, too complicated for this thread.

Access to care can probably be better solved through universal coverage, but not completely. I say not completely because even with universal coverage there needs to be X number of doctors to treat everyone. So even with universal coverage, you’ll get seen eventually but not always in a timely fashion, meaning the access problem is only imperfectly solved.

When you talk about lack of doctors that has little or nothing to do with the profits insurance companies make and nothing to do with whether or not we have universal healthcare. Lack of doctors is a function of doctor training and increases in doctor training relative to population.

Since the 1970s several things have happened:

  1. Demand for physicians per capita has increased. This is because more specialists have emerged, so single people have demand for more physicians in their lives than they did in 1970 when most people only needed a PCP.

  2. Population has increased substantially.

  3. The number of students graduating from medical school, in absolute terms, has not increased to keep up with population. What I mean is, we are not training as many doctors per capita as we were in the 1970s, and even in absolute numbers we aren’t training that many more doctors than we were in the 1980s. In fact we by and large quit opening new medical schools in the 1980s.

Supply of doctors and opening of new medical schools are controlled by a complex web of different things. On one hand you have the doctor’s “guild” the AMA, which has a vested interest in keeping the supply low because that causes wages to rapidly increase.

Unlike most other fields, the free market does not determine the supply of new doctors. This is primarily because the expense of training doctors is so high the market never really developed a good system for doing it. Instead we have a system where students take out a lot of debt (subsidized by government guarantees), and then enter residencies (which are expensive) that are mostly funded by Federal dollars. The Federal government spends over $10bn a year in compensating hospitals for medical residents, or around $110,000 per resident.

Opening new medical schools also involves a lot of government money. So the two big limiting factors to training new doctors (medical schools and residency spots) are heavily government controlled. The AMA also has a lot of power over medical schools, and a lot of political clout when it comes to influencing Congress to fund more or fewer medical residencies.

From the 1980s up until the early 2000s, the AMA successfully lobbied Congress to strictly limit new doctor production. For that reason in 2005 we were training in absolute numbers about the same number of doctors as we were in 1985, and the total population of doctors was even beginning to flat line.

The AMA wasn’t entirely malicious in this, they had reason to believe a “doctor glut” was coming. And that medical students would end up graduating medical school with $200k+ in debt and no job prospects. (A similar thing has actually happened to law students, employed lawyers still do well on the BLS compensation statistics but law school no longer guarantees a job practicing law and some law students have ended up with six figure debt and menial employment in non-law fields.) The doctor glut never came, because of an increase in demand unforeseen. Medical students do not go hungry when they graduate unable to find residencies.

To exacerbate this problem, at one point in time roughly corresponding to when baby boomers were medical school age we did have a surge in doctor training relative to population (this surge is part of the reason the AMA was afraid for awhile that a glut could overfill the market.) Some of these baby boomer doctors are now nearing retirement age, and because they represent an outsized portion of currently active doctors and most will be retiring around the same time, and because our per capita demand has gone up without corresponding increases in supply, we’ll be looking at a deep drop in active doctors that won’t be made up for by current production rates.

The AMA has recognized this and no longer actively lobbies to restrict the supply of doctors, and in fact is lobbying for more Federal money to go towards doctor training. But it takes 10 years to train a doctor and a long time to build a new medical school or even expand the facilities at extant medical schools. So even with political will and fiscal feasibility you can expect physicians to be in a state of undersupply for a very long time, and this is the source of the problems being described by the OP. (Spiraling healthcare costs and uninsured people is a separate, and mostly unrelated part of the healthcare crisis.)

Uh, right. :rolleyes: Not every medical condition is insanely complex. Quite a few can be diagnosed by Lesser Mortals if need be.

But, just to touch on the highlights:

  1. Increased chronic illness of lifestyle origin (obesity leading to diabetes and its knock on effects, as an example).
  2. Increased lifespan - when most people dropped dead in their 40’s we didn’t have a lot of 80 year olds to spend money and medicine one.
  3. Increased technology - in the old days, if your liver quit you just died, now we can, at least potentially, save your life with a liver transplant but a transplant and the subsequent absolutely vital long-term follow up care are WAY more expensive than simply dying. Likewise, we can now save accident victims who used to just die, but while some of these folks recovery fully others are left severely debilitated and require care for decades.

The solution here, short of training more doctors (which would be better) is “triage”. Of course, people are often uncomfortable with the notion. However, I think we have sufficient capacity that, if sanely applied (yeah, yeah, I know, incredible optimism there) everyone will get cared for, with the most urgent cases going first. Yes, that means you might have to wait six months or a year for hernia surgery if your condition isn’t particularly threatening. On the other hand, we might have fewer cases like this

Most people STILL only need a PCP, but they’ve been convinced otherwise. I recall a thread where people got very heated regarding who should do an annual pap smear, a GP or if sampling a few cells from a woman’s cervix really required a specialist, i.e. OB/GYN. Mechanically, it’s pretty freakin’ simple, really, and regardless of which doc does the sampling the cells are analyzed by a technician, not someone with an MD anyhow.

While the AMA does retain some influence, only about 40% of US docs belong to it nowadays, or something like that. They’ve lost a LOT of influence since the 1970’s.

This ties into loan forgiveness schemes - my college roommate had her entire medical school cost paid for by the Feds in exchange for being posted to whatever godforsaken remote outpost needed a GP for seven years. Well, some barren swath of New Mexico got an actual MD for their clinic and she was debt-free in seven years, and is still a GP (now in rural Wisconsin, which also needed a GP who didn’t mind the boondocks). That’s what I mean by “incentive”. Give people a guaranteed way to discharge their debt, you get medical coverage in undeserved areas, and you get another GP for the nation. Win-win, really. I’m sure a little thought could come up with other useful schemes of this sort.

I think you’re also forgetting that the supply of US doctors is not solely dependent on US medical schools - my last two GPs have been from Australia and India. Of course, that leaves the problem of the US vacuuming up qualified doctors which may well leave other places underserved.

Things which can easily be diagnosed by humans obviously can always be diagnosed by humans. But that does not mean it will be. This hits on a topic I can easily get off track on, but I basically am a believer that at some point in the far future so much of society will be automated and ran by machines that most humans will not be engaged in productive work.

The classical argument is technological advances put buggy whip makers out of business but create new jobs. That holds true to a point, but I’m a “believer” in an ultimate “singularity” of sorts when machine intelligence and physical ability far surpasses human ability to design ever better iterations of things. Once that happens I think very quickly most areas of the economy will no longer be human ran. Some people today can buy into that when it comes to things like manufacturing, but I took a very long view (500 years out) and I see even the white collar types who think their jobs would never be replaced by automation or the highly trained professionals like doctors and think they too could certainly be replaced by a highly advanced machine.

Once that happens there won’t be any human doctors to diagnose the simple stuff for one. Additionally, even if a human doctor could easily diagnose something, no one would trust that they wouldn’t miss a rarer and harder to diagnose disease that a machine diagnostician would reliably catch.

Loan forgiveness has virtually nothing to do with doctor supply. Large loans are a disincentive to becoming a doctor. However to get into a medical school you need almost a 4.0 GPA undergrad and a very high MCAT score. This is because vastly more people apply for each singular spot in medical school than is available, so the schools are ultra-selective.

When people talk about loan programs or incentive programs for teachers it is in response to a perceived lack of interest in majoring in education (I actually debate that is even a problem in general, but it’s beside the point.) Specifically we know we have some lack of teachers in certain subjects in certain areas. Since it’s easy to seat more education majors at tons of existing schools you can easily increase the total number of majors with good incentives.

It doesn’t really matter what financial aid incentives the government gives people considering medical school, the high salary and prestige associated with being a doctor have obviously created a huge demand for medical school spots above and beyond supply. That means the only way to actually increase the number of doctors is to increase capacity at medical schools and open new medical schools.

Loan forgiveness is a good tool for State & local government to reallocate doctors to areas lacking sufficient coverage. This is viable in part because in most States doctors collect themselves into the largest cities in the State predominantly because private practicing physicians in the various specialties make more money in those locations than they do in rural areas. So lots of States have highly effective programs that help pay medical school loans or etc to attract doctors away from areas that already have (relatively) sufficient numbers of doctors to areas with severe shortages. (You even see this with dentists as well, dental care is an important part of overall health and like M.D.s dentists also tend to not naturally go in large numbers into rural areas.) But that’s just shaking the pot up to more evenly distribute things in a desirable fashion, you aren’t actually increasing the contents of the pot this way–and cannot in fact. If there was a shortage of people applying to medical school then incentives could easily increase the number of physicians we train per year, but since we have more applicants than we can accept only increased bandwidth in the training system (medical schools + teaching hospital residencies) can increase the absolute number of doctors.

The healthcare crisis is caused by greed. Reducing the number of doctors will not help reduce costs. Any actual saving will be taken by the greedy. However computer diagnosis is on the way and already in use.

While expert systems IMO can and will eventually take over most of the medical system, they aren’t going to be solving any crises.

First of all, the crisis will be over, for better or worse, by the time the technology is available in enough quality and quality to make a difference.

Second of all because machine-based medicine will contribute to the biggest crisis we’ll face in the next couple hundred years – making human labor pretty much irrelevant to the functioning of a goods and services based economy.

Except that GP’s and family practice and some other areas where new doctors are in even more demand than, say, plastic surgeons, are NOT as high paying as people think they are, particularly in relation to costs like malpractice insurance and loan repayment. They aren’t necessarily an incentive to get more people into medical school, they’re an incentive to get people into the types of medicine that are most needed by society.

But a better distribution of service providers IS part of fixing the health care system. It’s not the total package, but it is a vital part of it.

Well at the extreme ends Technology can theoretically help hugely prevent unnecessary costs, but the best benefit would be from diagnosing improvements, rather than decision making. If You walk into the ER complaining of chest pains, and they can just whip out a tricorder and tell that you just have bad indigestion, you can save everybody a lot of work. Or hell maybe someday a smartphone will do it from your own living room.

One of the potential problems, cost-wise, could be the elimination of misdiagnoses. How many problems are dismissed as “indigestion” until death is inevitable? Perhaps successfully diagnosing and treating severe illnesses at an earlier stage would be more expensive than a late diagnosis followed by a quick death.

This reminds me of one of the arguments the tobacco industry lawyers made while trying to reduce their financial liability in the settlement – “Think how expensive all those smokers would be if they lived! We’d need more housing for them, they would be taking up jobs that would otherwise go to the next generation, …” Their claim was that killing so many of their customers created considerable benefits to society that weren’t being factored in.

Have you tried looking up medical expert systems? There’s a huge body of literature on the subject. My familiarity with the subject ended somewhere in the early 90’s, but at the time, they were being used for “House” levels of diagnosis – complex problems in identifying bacterial infections and so on.

Much of the work that your kindly GP does doesn’t require that kind of sophistication. Chances are, you already know what’s wrong with you when you show up and what you really need is a bit of poking and prodding and being told in a gentle authoritative voice that it’s no big deal and you should see if it gets better on its own.

However, I do recall that there was some research being done on expert systems to choose the least expensive/intrusive test that would provide information to differentiate between possible diagnoses. So to that extent, yes, there’s a role for expert systems to reduce the cost of health care.

Well. all we have to do is get rid of greed and the problem will be solved.
This is actually the opposite of the truth, if the problem of health care affordability can be solved it will be solved because of greed.
The answer to the OP is that automated diagnoses will be a part of any real solution but the powers that be will not allow them to be used in any significant way for a while. Doctors don’t want to replaced by machines any more than autoworkers did. The difference is that the AMA controls who gets to be a doctor and will ensure that there is always a shortage of doctors. They will also make rules making sure that only doctors will be allowed to diagnose pateints. So although a smart person with a checklist can diagnose people as well or better than most doctors it is illegal for them to do so. Because of this only doctors will be allowed to use the software, taking away most of the savings from the software.
The current path is unsustainable and at some point a process that is more automated and less doctor intensive will take over. It will be an expensive process to get there.

You wanna try and explain that?