IBM's Watson To Help Diagnose, Treat Cancer

Lucas123 writes "IBM's Jeopardy-playing supercomputer, Watson, will be turning its data compiling engine toward helping oncologists diagnose and treat cancer. According to IBM, the computer is being assembled in the Richmond, Va. data center of WellPoint, the country's largest Blue Cross, Blue Shield-based healthcare company. Physicians will be able to input a patient's symptoms and Watson will use data from a patient's electronic health record, insurance claims data, and worldwide clinical research to come up with both a diagnosis and treatment based on evidence-based medicine. 'If you think about the power of [combining] all our information along with all that comparative research and medical knowledge... that's what really creates this game changing capability for healthcare,' said Lori Beer, executive vice president of Enterprise Business Services at WellPoint."

Comments

What could possibly go wrong?

[idontgno]

We already have insurance case evaluators overriding a practitioner's medical judgments. Now, we'll have evaluators PLUS a very expensive rules engine* versus the overworked GP.

*And what, prithee, does the price of the system have to do with its credibility? Everything. If you sink a lot of money into something like this, you've already bet your money on whether it's right or not. No one is installing a Watson rig with an expensive data warehouse just for lulz, and no one's going to be able to casually second-guess this thing without massive evidence. It's going to be right all the time or BC-BS will look like a dope for spending so much.

Beside, it won Jeopardy! It must be right!

Re:What could possibly go wrong?

[robot256]

The nice thing is that its result is not just spat out of a black box--it gives a pretty accurate confidence measure, and actually links back to the articles that led it to each conclusion. That means the doctor can go and read them for himself. He may find articles he never would have found otherwise, and become a better doctor for it. It also gives the basis upon which to challenge the computer in court, if it comes to that ("Toronto", anyone?). I think the hope of WellPoint is that it will allow doctors to learn from research faster and more efficiently, so that young doctors learn faster and old doctors stay current. The more the doctors on the front lines of medicine know, the better patient outcomes will be. Nobody is saying the computer is going to ever replace the doctors altogether.

Then the legal department will come and screw everything up, of course, but we can wait a little while before that happens.

Re:What could possibly go wrong?

[Trepidity]

Computers are actually better at certain kinds of diagnosis than the overworked GP, though, and have been for years. In particular, computers are very good at conditional probability, and at combining information from thousands of study results that a typical GP doesn't have time to keep up with. The MYCIN AI system beat most doctors in diagnosing blood infections over 30 years ago, but wasn't adopted in actual medical practice mainly for political reasons.

Re:What could possibly go wrong?

[Dunbal]

Yeah - "helping oncologists diagnose and treat cancer" my ass. More like: This patient has a very high probability of having cancer - drop his coverage, quick! Oh and after we cancel his insurance, tell him to get checked for cancer.

Re:What could possibly go wrong?

[ColdWetDog]

The nice thing is that its result is not just spat out of a black box--it gives a pretty accurate confidence measure, and actually links back to the articles that led it to each conclusion.

A question and a comment.

First, how do you know this? It really didn't say how it is going to get an 'accurate confidence measure'? Which leads me to my comment - the completely untested assumption here is that the answer lies in the current literature, if only you could wade through it. That isn't at all clear to me. Much of the 'best' data comes from double blinded placebo controlled studies. The big problem here is that the patients are typically carefully selected for having as few other co morbidities as possible. That's very useful from a research situation but makes the data poorly generalizable to the average patient on the street who has no particular interest in being 'simple' (or rational or compliant). The rest of the medical literature is basically crap. "Expert" opinion which turns out to be wrong as often as not. Observational studies which almost always inflate the efficacy of treatments and can only provide correlation. Much of medicine has really never been studied carefully at all.

Next, if they're using any insurance company's billing data, well you might just as well consult rabbit entrails. In the US the vast majority of that data is entered in an obfuscated, outdated and thoroughly whimsical system called ICD-9 (International Classification of Diseases, version 9). The REST of the world with the possible exception of North Korea is on 10 - a system that is much more useful. But even using one of the more sophisticated medical database systems is still unlikely to give you the detail you need to actually treat someone.

Of course, there is little useful information on how this will work - whether the doc will consult this wonderful oracle or if the insurance company will send you a form letter six months after the patient died saying you should have done something different. If they are going to go through with this,, I hope to hell they are going to carefully monitor it's success (or lack thereof) over time. And do that honestly. My money is that it won't help all that much.

Re:What could possibly go wrong?

[elsurexiste]

Wikipedia suggests that the real reason was technical: loading the required knowledge into the system was painful. This is more related to my own experience: ESs don't get adopted because UI is crappy at bests.

Re:What could possibly go wrong?

[Oxford_Comma_Lover]

> your life literally depends on who you happen to get lucky enough to see.

This is why for anything serious, you are very careful about who you see. It's really a combination of luck and networking to reliable people. The hardest part (for a slashdotter) is finding one person who is good enough and knows enough that they will send you to what is probably the right place. Someone can have a wonderful reputation and be a bad surgeon, and sorting wheat from chaff is really, really hard, especially when the medical community is tight-lipped to begin with, and it's not like they provide patients with empirical information.

Re:What could possibly go wrong?

[rjstanford]

Here's a joke for you:

What do they call the person who graduates last from Medical School?

Doctor.

As a patient, I'd like to know what sources my provider has consulted and based a diagnosis on. I don't want my health dependant on which PA or GP I get when I go to the doctors office or to the hospital. Doctors may catch the obvious things, but when its not obvious, your life literally depends on who you happen to get lucky enough to see. That's a pretty sad state of affairs, and its great if this helps.

Here's some truth for you. There are ~310 million people in the US alone. There is one "best oncologist." Most of the people are never going to see him or her.

The good news, and some more truth, is that most of the people - even those with cancer - don't need to.

The bad news is that many of them will think that they do and bitch and moan when they don't.

Re:What could possibly go wrong?

[nbauman]

Do you know any oncologists?

I've gone to lectures, and talked to a lot of them. I've heard that there are dumb oncologists, but I've never met one.

Oncologists are pretty smart. They *already* know the medical literature cold. There aren't that many high-quality clinical studies on a major cancer -- say, breast cancer -- and an oncologist who treats breast cancer will know them all. They may not have *read* them all, because they've heard the results of the studies when they were first reported at medical meetings. But they know them, and they understand the biological mechanisms of cancer as well as anybody knows.

And oncologists *already* use computers in those applications where it's useful. They have electronic medical records in at least the major hospitals. (Some doctors complain that EMRs make it too easy to fill up a medical record with irrelevant information that nobody has time to get through. The old paper records forced doctors to get to the point and concentrate on relevant information.)

PubMed http://www.ncbi.nlm.nih.gov/sites/entrez?db=pubmed is free and probably the best medical database created for any money.

Oncologists aren't ashamed to search Google when they hit a dead end.

Peter Norvig wrote some great algorithms for Google, and they can do some wonderful things -- translate languages, finish your search request, and other amazing things.

Norvig was lucky. The kind of problems that doctors face haven't fallen to clever algorithms yet.

2 points:

(1) The best way to manage medical information is in a doctor's head. If computers can help, that's great, but don't get overconfident.

(2) You'll never know if a particular computer application works in medicine until somebody does a well-designed randomized controlled trial to find out -- the old scientific method. The results of computers in medicine are mixed. Sometimes it works, sometimes it doesn't. If your life is important, use the scientific method.

Re:What could possibly go wrong?

[xyourfacekillerx]

The nice thing is that its result is not just spat out of a black box--it gives a pretty accurate confidence measure, and actually links back to the articles that led it to each conclusion.

I'm trying to imagine the scenarios in which a doctor would be forced to resort to using a sophisticated computer to diagnose a patient with cancer, but if he doesn't have the time to refer to a good oncologist, what makes you think the doctor has the time, or is even qualified to understand, the articles in question, so that by referring to them personally the doctor could confirm the diagnosis? Look, I KNOW my family practitioner doesn't have the specialized training to catch cancer. I KNOW that reading articles and reviewing billing practices (a terrible metric if you think about it) would lend any sort of qualified expertise. So, I really, really have doubts that a computer database - no matter how spectacular the algorithm behind the "information compiling" may be - is a better resource than my ignorant family practitioner.

All it takes is one false negative in a case where an oncologist could have made a positive diagnosis, and it proves relying on the computer is unsafe. Are they going to put it through 900 rigorous trials like I just described? Would you trust your wife or daughter's illness to Watson if they don't? Even if they do?

Long time coming...

[understress]

Having had to deal with the medical world for the last few years with doctors trying to determine what is wrong with my wife (still no definitive answers yet), and have them treat her, I am shocked that something like this hasn't been done before.

Even a simple db that cross references diseases to symptoms / blood work results (and other test results) doesn't seem to exist. It's 2011, you'd think that doctors could order up a set of tests based on their initial thoughts, input the results to a program, and have the program guide them with possibilities to try and narrow down the search of what may be wrong. The symptoms that my wife has can be linked to MANY different diseases, but in the end, each disease has something that makes it unique. It should be a simple path of elimination. Test until you find the one disease that fits that persons set of results.

I'm not a medical doctor, but I've done a lot of research on the web about what is wrong with my wife (yes I know it's not all correct) and I'm shocked that twice now I've had to ask the doctor to perform some tests and find out if a certain condition exists, and it did. Simple, her symptoms are this, these blood tests could tell you yes or no if that's what you have.

Doctors seem to want to just prescribe something that should help the symptoms. How about we figure out what is wrong first and then treat appropriately?

I'm glad to see something like this finally being developed. Like I said, it's 2011, some of the ways things are these days is just crazy considering the computing power we have (personal, national, worldwide) available to us.

Re:Long time coming...

[PRMan]

Consciously or subconsciously, most doctors make more money by treating symptoms repeatedly than by curing people. All-in-one systems like Kaiser would be the exception, since they own the hospitals, insurance plan and everything, they are incentivized to actually cure you so you stop costing them so much money!!!

This is a big problem in medical and dental insurance in this country.

Re:Long time coming...

[sconeu]

I sympathize with you, understress. My wife and I went through 18 months of hell, with tons of red herrings, including a dismissal -- "Oh, you're just suffering from empty nest. It's psychosomatic." -- before she was diagnosed with ALS.

The problem with rare diseases is the old saying, "When you hear hoofbeats, look for horses, not zebras."

Re:Long time coming...

[grmoc]

FYI, it has been done before. The computer did better than the GPs (and this was decades ago), however, noone was wanting to be the liable party.. and so, it never saw real use outside the study.

Score some more "benefit" for lawyers and the people who litigate.

Re:Long time coming...

[jd]

Identification keys are commonplace in most fields. I wouldn't personally use something like Watson for that, though. I also agree that there are some amazing gaps in the medical databases that are around and that medicine should be more about the underlying mechanisms and less about the symptoms. (If you use thick enough paint, you can hide the cracks in a house that's subsiding. It won't stop the house collapsing, though.) The pressure for evidence-based medicine might help, but again that depends on what is considered evidence.

What is wrong is, however, a bit tougher. It used to be thought that in genetic conditions that one condition equalled one gene. Turns out that one condition can equal certain combinations of maybe a couple of hundred genes but not other combinations and even if all the genes are present in the diseased form, you still need the epigenome to be a certain way and that depends on the environment of not only the present but also the past 2 generations. In those cases, knowing what is wrong depends on quite a lot of information that is very difficult to obtain.

That's obviously not all illnesses, though. Viruses and bacteria should be much easier to directly observe and therefore directly treat. (I dislike intensely indirect observations via the immune response, as that only works if the immune response was correct in the first place AND if the observation of the response is valid. No matter how probable something is, if you stack enough probabilities together you end up with something exceedingly unlikely. If you're capable of direct observation, the probabalistic route makes no sense. And even the NHS can afford decent microscopes.)

Re:Long time coming...

[clem.dickey]

An anecdote from Dr. James C. Cain (former head of section, gastroenterology and internal medicine at Mayo), from about 1981:

A patient came to May Clinic with vague symptoms. One histologist remarked "This guy has weird blood. I've seen it before, but can't remember where." Several days later the histologist came back with the book where he had seen that "weird blood." Leprosy. Mayo didn't get many lepers.

"We were just lucky," said Dr. Cain, "that the histologist remembered the pattern. But imagine what we could do with a computerized search."

Re:'idocracy'

[electron sponge]

Who tagged this 'idocracy'? Is that even a word?

If this were an Apple medicine machine, maybe it would be iDocracy.

Idiocracy

Tag whoosh?

Re:Diagnosis.......Complete

[electron sponge]

It seems you have a severe case of being a little pussy, I am prescribing that you man the fuck up.

Doctor: But Watson, the patient is a 5-year old girl!

Re:Then why waste money on cancer treatments?

[robot256]

That brings up a good point though. If the computer says it's a bad idea, and backs that up with evidence, might it not actually be a bad idea? The whole point of evidence-based medicine is to improve outcomes (and save money) by reducing late diagnosis, misdiagnosis and unnecessary procedures. If, in fact, the computer is right most of the time and you don't actually need that procedure, then it will save money. There will always be errors, whether it's the fault of a doctor or the computer; the goal is to reduce their cost and frequency.

Here's another take: If you assume a certain amount (or even most) unnecessary procedures are a result of defensive medicine and doctors covering there asses, then might not the computer give them an excuse to omit those procedures which, medically, they already know are unnecessary?

Re:Tumor?

[jamiesan]

Would you like to play a game? How about Global Thermo Nuclear War^H^H^H^H^H^H^H^H^H^H^H^H^H^HJeopardy?

Hold on, this might actually be a good idea

[plurgid]

As someone who recently was stuck in our incredibly broken medical system with cancer that was mimicking symptoms of other diseases (which were coincidentally much more profitable to sell "management" drugs for), I actually think this could work.

You guys are on the tip where you're thinking a cold heartless machine will be making the rules, like it's a bad thing.
Look, I was stuck in a small town where the biggest industries are defense contracting and medical services. Do the math. As long as my symptoms looked plausibly like something that was going to make everyone a lot of money to sell treatment for, there was no F-ing WAY anyone was going to have any shred of curiosity about what the real problem was.

It's not that people were being dicks. They were being human. Nobody WANTED me to continue to get sicker, but nobody at the levels low enough to notice knew any better, and the people high enough up the chain to know better were too busy counting their money and running the small-business that was their practice to notice.

In the end it was ME who had to hit google, find a research university, verify that they were covered by my insurance, and basically go to my doctor and stage a sit-in until the motherfucker wrote me a referral. That shit SAVED MY LIFE.

And I'll say it again. Googling my symptoms and having the self confidence to question the system because I KNEW something didn't add up SAVED MY LIFE.

My insurance was buying the equivalent of a mid-size sedan on my behalf for medications for a disease that I did not have (that in the end were indeed making me much sicker). I'd bet Watson would have picked that shit up pronto and forwarded me up the diagnostic chain.

It has the potential for abuse, sure. But I actually would rather trust a correlation engine to pick shit like that up than a bunch of self-interested medical professionals cum-entrepreneurs. Believe that.

Re:Diagnosis.......Complete

[Dunbal]

It seems you have a severe case of being a little pussy, I am prescribing that you man the fuck up.

Doctor: But Watson, the patient is a 5-year old girl!

Not after all that testosterone she won't be.

Re:Given the choice

[Dunbal]

Ingested vitamin C cannot exceed a certain level in the blood stream

Of course not, it's water soluble and filtered out in the glomerulus along with all the other water-soluble stuff in your blood which includes every other vitamin except A,D,E, and K which are fat soluble. Then the kidney reabsorbs the water soluble stuff it needs, like glucose, vitamin C, etc through sodium dependent transporters. However like all enzymes, these reabsorbtion mechanisms are saturable. So no matter how concentrated the Vitamin C in the ultrafiltrate, there is a limit to the rate of re-absorbtion.

The blood level doesn't change.

[citation needed] If the blood level of Vitamin C is fixed and never changes, explain scurvy. Oh, so what did you mean by "it never changes" then? You mean it's possible to go from near zero to a maximum limit? OK yeah I'll grant you that. The maximum plasma concentration is set by the kidney, like I described previously. You could inject yourself with as much vitamin C as you like and, assuming you survived, you would just piss it all out. The rate of glomerular filtration is quite impressive when you count it in liters/hour.

merely that he succeeded in producing a proof of the total lack of toxicity amongst healthy cells

You are confusing cells in a Petri dish with a living, breathing multicellular organism.

While Pauling might have had his brilliant moments and certainly contributed to science, this does not mean that everything coming out of his mouth is a golden nugget of wisdom granted by the gods. The whole Vitamin C thing is quackery and there is no evidence that it does anything to help with common colds or influenza. It's involved primarily in collagen synthesis which might be useful for the burns patient, but not really in the influenza patient - unless you consider pulmonary fibrosis a successful outcome.

Oh, by the way I happen to know a little about how the human body works, did you notice? Go ahead, keep arguing. "I have studied it and you have not" - Isaac Newton.

Meanwhile,

[optymizer]

in my country:

• - we bribe the doctors. If you don't bribe them, they don't treat you and most won't even diagnose you. It's because "their salaries are low".
• - most doctors drink their way through medical school.
• - many doctors fail their exams again and again and attempt to buy the medical degree (and those who have enough money succeed in buying it).
• - lie their asses off to get your money and then prescribe a ton of drugs, more than you need, just to make the pharmacies happy.
• - do _not_ stay up to date with the latest in medical research.
• - there are a handful of good, honest doctors who want to treat people and are forced to do magic with the few antiquated tools they have. Access to the only MRI scanner in the country, which also happens to be the most advanced piece of medical technology we have, is controlled by the private clinic that owns it and ... no, you're not getting scanned unless you're the president's son or you're filthy rich.

I'll take Watson's diagnosis any day of the week!

virtual doc you get leprosy!

[Joe_Dragon]

Lisa: Maybe I ought to check with the doctor.
                [Lisa, Bart, and Homer gather around Lisa's
                computer. She starts a program that displays a
                medical logo -- the one with two snakes wrapped
                around a staff]
Snake 1: Welcome to "Virtual Doctor."
Snake 2: From the makers of "Dragon Quest," and
                "SimSandwich."
Snakes 1 + 2: Enter symptoms now.
Lisa: Let's see. [types on keyboard] Crusty sores?
Homer: Yes.
Lisa: Horrible wailing?
Homer: Yes, yes!
Lisa: Any exposure to unsanitary conditions?
Bart: Duh! We're pigs.
Lisa: [finishes typing] Okay. And ... diagnose. [pushes
                a key]
Virtual Doc: You've got: leprosy.
Homer +
  Bart: Leprosy?! Aaah! [point at one another] Unclean!
Bart: Unclean!
Homer: Unclean! Help us virtual Doc! Look at me -- I'm on
                my knees.
Virtual Doc: Goodbye. [leaves the virtual office]
                [Homer and Bart whimper]

Re:Limited treatment on the way

[PopeRatzo]

If it isn't obvious, this is tiered medical care at its finest. Sorry! You don't make enough income, and your dependents, upon your immenent death, won't be able to foot your lingering bills over the next decade. You get treatment plan B3 instead.

Of course, you're right. To the extent that any insurance company, including (especially) Blue Cross Blue Shield, is using the power of supercomputers, it's not to diagnose and treat disease, but to figure out more creative ways to NOT PAY for patients' treatment.

Let's not forget that the entire business model of health insurance companies is based on paying less for customer care than those customers paid into their policies. It's based on denying coverage. Think about that: the whole point of health insurance is to not treat patients, not treat disease.

This is why it's just insane to allow for-profit corporations to be involved at any level of health care, including pharmaceuticals. Think of all the money that is spent on researching new drugs. The entire thing could be paid for with public funds and the drugs could be made public domain and it would cost less than George Bush's Medicare Part D which was nothing but a trillion-dollar giveaway to the transnational pharmaceutical companies.

As long as we follow the for-profit model of health care, things are only going to get worse, and "medicine" is going to lead us to some very dark places. It already has.

subject

[Legion303]

...and then Watson will help the insurance company decide how best to deny coverage for those very same symptoms. Truly a marvel of the technological era!

powers for good, or evil?

[BenBoy]

Physicians will be able to input a patient's symptoms and Watson will use data from a patient's electronic health record, insurance claims data, and worldwide clinical research to come up with both a diagnosis and treatment based on evidence-based medicine

Then, the system cancel the patient's policy millions of times faster and more accurately than humans doing the same job might.

Re:Limited treatment on the way

[trout007]

Unfortunate you are correct with everything except for calling what we have in the US insurance. Insurance in all other areas of life is a way to pay a little cost up front in order to be spared the expense of a rare but costly event in the future. Examples?

You get homeowners insurance in case of a fire, burglary, storm damage. These events don't happen often. I've paid homeowners for 12 years with no claims.

You get car insurance to protect for accidents, liability in an accident, or having the car stolen. These are usually higher risks than homeowners insurance so the cost is more vs what is actually being insured.

Term life insurance. Pays whomever you want upon your death. Very cheap for the young and it gets much more expensive as you get old.

Health "Insurance" is no such thing. I use healthcare all the time. Weekly if you count my family. It is not insurance is is some sort or prepaid health service contract.

Whats the difference? All of the former insurance I don't want to use. I don't want my house to burn down, I don't want to get in a car accident, I don't want to die. The insurance company and I are on the same page.

But with healthcare if I'm paying for the service I'm going to use it. Everyone has aches and pains and sniffles. If you actually had to lay a doctor what it costs out of pocket you would take more care as to how you spent those dollars. Also doctors are to blame for having a monopoly on prescribing medicine. This forces what should be a 10 minute $4 trip to the pharmacy into a $80 afternoon.

If you eliminate doctors monopoly on drugs and go back to real insurance to cover things you can't pay for out of pocket you would see costs drop.

Re:mixing up citys = a very long distance call is

[Rich0]

Well, this is why a doctor is still on the hook to interpret the results.

Most doctors will probably just ignore it. In my experience the average doctor has the average treatment they're comfortable with when a patient comes in with a given set of circumstances. They get paid the same for the visit whether it takes 5 minutes or 50 minutes, and 99.999% of the time ruling out a rare ribosome disorder is time wasted. For 99.999% of the population that works out just fine, but if you happen to have a rare ribosomal disorder you're pretty much out of luck as every doctor you turn to will first start treating you for some common malady that is hard to diagnostically confirm.

In the real world, House would be driving a used car, even if he could miraculously diagnose problems correctly (which of course is unrealistic to begin with). He just wouldn't have enough patients to bill. The money is in extracting $100/day from every other patient in the hospital regardless of outcome.

Profit motive is obvious

[Quila]

But you can tone down the cynicism.

Misdiagnosis costs the insurance companies billions a year. An early-caught cancer costs much less to treat. Performing unfruitful tests in search of difficult diagnoses costs billions more.

If this produces faster, more accurate diagnoses it could save them billions every year.

It does not have to be a zero-sum game where the patient has to lose if the insurance company is to win.