Watson got its first job! IBM has been looking for a way to put the supercomputer system to work ever since it beat Jeopardy champions Ken Jennings and Brad Rutter back in February 2011 (for more background information on Watson and the Jeopardy face-off, see Christine Hawes’s June 23, 2010 blog post). Following through on its initial target for use in the medical industry, Watson will now be working with Well Point Inc., the nation’s largest health insurance company (based on enrollment figures), which operates Blue Cross Blue Shield plans for fourteen states. Watson will eventually help the insurance giant choose treatment options and medicines for patients, but first it has to bone up on its medical knowledge, absorbing a library of medical textbooks and journals along with WellPoint’s history of approved medications and treatments. Then, once a doctor verbally tells Watson about a patient’s medical history and symptoms, the system will sift through its vast universe of information to come up with a list of the most likely diseases for diagnosis, ranked in order of its confidence in the answers. Once it is up and running, doctors should be able to access the application from their computers or hand-held devices to assess their patients’ complaints.

While some are praising this new advance in technology as a way for doctors to get a handle on the ever-growing wealth of medical information, others are wary about the potential abuses of this new use for the system. Initial tests have shown that Watson comes up with obscure diseases and can drastically change its projected diagnoses the more information it gets about a patient, which can lead doctors all over the map as they try to figure out their patients’ ailments. In addition, many doctors already use similar programs, such as Isabel or UpToDate, to come up with diagnoses and treatment options. And with the recent debut of Apple’s Siri program in its iPhone 4S, it seems that other voice-activated search programs will soon be on the horizon and could be applied to the medical field.

More worrisome is the potential for the insurance company to deny benefits or restrict approved treatments or medications based on what Watson says. This possible abuse has already proven to be a problem with various insurance companies’ use of other computer systems for both health and property claims, strategically undervaluing claims to maximize profit for the company and leaving the insured to foot the remainder of the bill. Additionally, there is the risk that someone could hack into the centralized computer system and expose people’s private medical information, which is not only used for treatment of that individual but also stored to help in diagnosing others. As the insurance field becomes more technologically advanced, it appears that it will also require greater oversight so that we can reap the benefits of the system while avoiding the risks. Until such protections are put in place, it seems that not everyone will “welcome our new computer overlords.”

– Kendall Short
Image Source

Tagged with:
 

11 Responses to Elementary, My Dear Watson

  1. Andrew Farrell says:

    Many hope that Watson will help to decrease the per capita cost of healthcare in the US and increase overall care quality. However, the chances of that happening are still undetermined. Doctors presented with a wide range of diagnoses could react by intuitively using Watson’s diagnoses in conjunction with their own judgment to more efficiently treat patients. On the other hand, doctors could view Watson’s ranked diagnoses as presenting more work. They could order more wasteful tests because Watson’s predictions make them think twice about their own diagnoses and prompt them to take precautionary measures against future malpractice suits.

    Watson may also fail to be as useful as some hope it will be. Some scholars have identified two types of chronic diseases: intuitive chronic diseases and rules-based chronic diseases. Intuitive chronic diseases like back pain and schizophrenia do not entail obvious diagnoses or courses of treatment. Thus human judgment and experience still plays a large part in their resolution. Rules-based chronic diseases like diabetes are more susceptible to treatment guided by evidence-based medicine. Computers are more capable to handle these conditions.

    The interesting thing to watch in the future will be Watson’s ability to use its database of knowledge to treat intuitive chronic diseases. If it can do so effectively, and doctors are able to use its suggestions efficiently, then healthcare costs may be contained and healthcare quality increased.

  2. Susan Reilly says:

    The insurance implications of this technology will be interesting – especially in terms of procedures covered by insurance as a “medical necessity.” If the computer program is making the diagnosis instead of the doctor himself, will this change the definition of “medical necessity” and how it is determined?

  3. Collins Kilgore says:

    I am certain that one day, probably not too far off, there will be computers that make better diagnosticians than humans. Watson, as it is now, is not that computer. It sounds like what this application is meant to do is to play an assistant’s role, giving doctors a range of possible diagnoses for a set of symptoms. It’s pretty good marketing and may even be good medicine. What I wonder about is when computers do, in fact, become better at identifying diseases, how this will play out in the medical community and in the law. As Ryan pointed out, there are huge ramifications for malpractice issues. Doctors may be at a greater risk for malpractice claims if they do not use computers–or don’t follow the computer’s advice–in making diagnoses. Doctors who use the computers would likely have lower malpractice insurance costs. And, if computers are proven to be better at diagnosing illnesses, might it not spell the end for misdiagnosis malpractice claims? That is, even when the computer is wrong, it would still be more prudent for treating doctors to follow its advice. And, wouldn’t this justify an insurance company’s denial of coverage for any treatments that weren’t within the scope of the computer’s diagnosis? All of this feels a little wrong, but it might actually be better medicine for the population as a whole. I think a just system would be for insurers to set aside money for those times when the computer is wrong. It would look like the current malpractice system, except the air of culpability would be taken off the shoulders of the doctor and shifted to the machine.

  4. Caitlin Angelette says:

    It seems like Watson and the database that would feed him might be our best hope for evidence based medicine, but evidence based medicine is not necessarily the panacea that it appears to be.

    Something that comes up a lot in Health care law is that there is this aspect of the professional paradigm that there is such a thing as a scientifically indicated correct response to a medical issue. This idea has been more or less undermined by some work done in the 90s called the Dartmouth Atlas, which showed that doctors vary widely in their approaches to treatment of the same illness.

    However, they haven’t given up on evidence based medicine, and I would be curious to see what a real database that would allow evidence based medicine to come up organically (with or without Watson’s input) would do to the treatment variation. The Dartmouth Atlas study suggested that the variation results from various training. If doctors were given access to a database that collected patient symptoms and outcomes from across the country, would they alter their practice style to accommodate evidence that their technique is not necessarily the best?

  5. Henson Millsap says:

    This fits within the larger push towards computerized medical records over the last decade, which I agree is long overdue, but it raises huge privacy concerns as you mentioned. After the public outrage over airport body scans, I can only imagine how people might react if one of these online databases is to found to have accidentally leaked, or worse yet sold, their medical records to a third party. Like all other emerging technologies this will obviously require careful regulation.

  6. Katie Kuhn says:

    I’ll be curious to see if using Watson as a diagnostic tool becomes a common practice for physicians. After all, specialists and general practitioners alike may face few cases that require trolling through medical information to find a “probable” diagnosis that is otherwise evading them. But when that need arises, Watson has the potential to act as a well-versed and helpfully suggestive sounding board. The utility of Watson may lie less in the list of possible diagnosis it produces and more in the explanation underlying the confidence intervals. Understanding how Watson weighed different factors in generating conclusions could provide doctors valuable assistance in evaluating the salience of symptoms when diagnosing and treating patients.

  7. Andrew Harline says:

    While I am all for advances in technology, this story leaves me feeling leery. My biggest fear is that, because a machine can do the work for them, human doctors will eventually become dis-incented from learning the information themselves. What incentive is there for someone to work hard and learn if he is just going to be told the answer by a machine? I saw Watson on Jeopardy. While I was pretty impressed when he got a correct answer, I was also blown away by how wrong he was on other answers. If human doctors become dis-incented from learning their stuff, I worry there will be nothing to filter Watson’s incorrect medical diagnoses.

  8. Ryan Sawyer says:

    This seems like it could be a useful tool in expediting the process for providing patients with an accurate diagnosis. Outside of the inherent problems with insurance providers, I am curious how medical malpractice suits would unfold. Would the target of malpractice suits be shifted to Well Point Inc.? Can the use of Well Point software provide an affirmative defense for doctors?

    I would also be interested in the feelings of the medical school academic community. If a computer can accurately determine a patients diagnosis based on symptoms, will this alter the curricula of medical schools? Rather than focusing on memorization, could a medical education become focused on the use of technology?

  9. Raymond Rufat says:

    I welcome this technology in the medical arena. I think it is important however that patients are made aware of the factors that go into the computer’s diagnosis. This kind of information would be important to patients decision making in deciding to go through with certain kinds of treatments or selecting health care plans. Furthermore, if the technology were to be released to the public it would be very helpful for patients to make cost efficient treatment decisions, in the way that WebMD already facilitates this.

  10. Cal Albritton says:

    I for one welcome our computer overlords. With so much more information available instantaneously, I’d love a doctor to use this on me. I definitely understand to concerns though; the computer can’t have the final say. But as a patient, I think I’d like to get a chance to see all the info, even the unlikely outliers. Maybe it’s just me embracing my inner hypochondriac.

  11. Alexandra Pichette says:

    The potential for abuse by the insurance companies seems huge. However, I wonder how much insurance companies already work off a system like this, albeit a simpler one; such as using a rubric and pre-approved and pre-denied treatments. Maybe the biggest potential difference will be how much the insurance companies use this new system to disregard the opinion of Doctors- who can rely on the human factors Watson does not possess, like intuition, experience, and the capacity to recognize outliers.