After defeating all comers on the television quiz show Jeopardy!, Watson -- IBM's artificial intelligence system -- is being put to work in healthcare. The first medical application of its massive computational powers will be to aid in diagnosis for the insurance company WellPoint. In a recent discussion on Medscape's Physician Connect, an all-physician discussion group, this new use for Watson -- and its implications for the future of medicine -- were vigorously debated.
Many physicians saw Watson's incursion into medicine as misguided and doomed to failure; others saw Watson as a useful tool to assist in diagnoses. A small and colorful minority worried about a robot takeover of medicine -- and the planet.
An internist, despite acknowledging Watson's ability at games, claimed that the demands of medical diagnosis are much more complex: "Diagnosis and medication of any individual patient start with at least dozens to scores of first-level branches and dozens to hundreds of covariables. Watson is a simple, albeit very fast, machine, but it is absolutely ignorant by definition. Putting a turbocharged V-12 engine on a lawnmower will not turn your lawn into a botanical garden."
A surgeon questioned how the information Watson would analyze could be obtained without the assistance of a doctor: "Who will put in the information? I did a preprinted patient questionnaire to elicit answers related to certain symptoms and believe you me, you need human input to help patients come up with answers."
A neurologist brought up the ineffable qualities that only a human can possess: "These computer algorithms, I think, will not take into account the intuition factor, which can be described in different ways but is a human trait and, when combined with intelligence, knowledge, and observation, is very useful."
But some physicians dismissed these misgivings and welcomed any assistance Watson could provide in the future.
"Let's not exaggerate all this. Computers are meant as a tool to assist us, to justify our intuition and facilitate diagnosis and treatment. In the early '90s during training, without UpToDate, MDConsult, etc., we resorted to reviewing studies we got on MEDLINE. Shortly thereafter, computer access on the floor and then broadband availability allowed us to make sane and cutting-edge decisions," wrote one internist.
"Having quick access to truckloads of data can help you arrive at quicker diagnos[e]s. Truckloads of data coupled with good computer reasoning can help a doctor stumble upon diagnos[e]s that he hadn't considered," added a pediatrician.
An emergency department doctor joined the supporters by writing, "If well programmed, Watson, er... Dr. Watson that is, could help guide us to consider those things we rarely see and consider. It's akin to auto-pilot: great to help fly big complex aircraft."
But a surgeon quickly countered the analogy, stating: "Recent deadly crashes have been attributed to pilot overreliance on the auto-pilot or misinterpretation of auto-pilot indicators and misunderstanding malfunctions."
Though the potential applications of Watson in medicine are broad, some physicians concentrated on the issue immediately at hand: the fact that the first use of Watson in healthcare is to aid an insurance company.
A worried pain management specialist asked: "World's first thinking computer's first task is to help a medical insurance company? ... I think we all know where that's going."
Several physicians joked about the increased speed at which Watson would be able to deny claims.
An internist welcomed Watson with weary resignation: "If [insurers] use it, all I have to do is document what the [insurers] recommend and tell the patients to call them back should it fail. I don't fight these anymore; I tell patients that that is their [insurer's] decision, not mine."
A general practitioner, supportive of the machine, feared the tasks its masters at WellPoint would use it for: "If I could have Watson at my desk, I'd be happy. It takes a lot of time to look through info on eMedicine, UpToDate, etc., depending on how familiar I am with the information to start with. But putting it to use for an insurance company -- bad idea."
Some took it a step further, suggesting (with various degrees of irony) that Watson signals the beginning of the end for humans. Many invoked science fiction films such as 2001 and The Terminator to demonstrate the dangers of artificial intelligence.
A surgeon summed it up with a dire forecast: "When the computers have it all figured out and can reproduce then they won't need us."
Some physicians sat somewhere in the middle of this debate, recognizing both the danger and the rewards that Watson may bring. "It can be a diagnostic aid for physicians; unfortunately, my cynical side tells me it will be marketed as a replacement for the personal face-to-face visit between a physician and patient," hedged one surgeon.
The effectiveness and the consequences of Watson are likely to be debated and analyzed as they continue to develop. But the use of powerful information technology seems likely to become a permanent part of medical practice.
The full discussion of this topic is available at: http://boards.medscape.com/forums/.2a0d8d05/37. Note, this is open to physicians only.
 |
An internist, despite acknowledging Watson's ability at games, claimed that the demands of medical diagnosis are much more complex: "Diagnosis and medication of any individual patient start with at least dozens to scores of first-level branches and dozens to hundreds of covariables. Watson is a simple, albeit very fast, machine, but it is absolutely ignorant by definition. Putting a turbocharged V-12 engine on a lawnmower will not turn your lawn into a botanical garden."
A surgeon questioned how the information Watson would analyze could be obtained without the assistance of a doctor: "Who will put in the information? I did a preprinted patient questionnaire to elicit answers related to certain symptoms and believe you me, you need human input to help patients come up with answers."
A neurologist brought up the ineffable qualities that only a human can possess: "These computer algorithms, I think, will not take into account the intuition factor, which can be described in different ways but is a human trait and, when combined with intelligence, knowledge, and observation, is very useful."
But some physicians dismissed these misgivings and welcomed any assistance Watson could provide in the future.
"Let's not exaggerate all this. Computers are meant as a tool to assist us, to justify our intuition and facilitate diagnosis and treatment. In the early '90s during training, without UpToDate, MDConsult, etc., we resorted to reviewing studies we got on MEDLINE. Shortly thereafter, computer access on the floor and then broadband availability allowed us to make sane and cutting-edge decisions," wrote one internist.
"Having quick access to truckloads of data can help you arrive at quicker diagnos[e]s. Truckloads of data coupled with good computer reasoning can help a doctor stumble upon diagnos[e]s that he hadn't considered," added a pediatrician.
An emergency department doctor joined the supporters by writing, "If well programmed, Watson, er... Dr. Watson that is, could help guide us to consider those things we rarely see and consider. It's akin to auto-pilot: great to help fly big complex aircraft."
But a surgeon quickly countered the analogy, stating: "Recent deadly crashes have been attributed to pilot overreliance on the auto-pilot or misinterpretation of auto-pilot indicators and misunderstanding malfunctions."
Though the potential applications of Watson in medicine are broad, some physicians concentrated on the issue immediately at hand: the fact that the first use of Watson in healthcare is to aid an insurance company.
A worried pain management specialist asked: "World's first thinking computer's first task is to help a medical insurance company? ... I think we all know where that's going."
Several physicians joked about the increased speed at which Watson would be able to deny claims.
An internist welcomed Watson with weary resignation: "If [insurers] use it, all I have to do is document what the [insurers] recommend and tell the patients to call them back should it fail. I don't fight these anymore; I tell patients that that is their [insurer's] decision, not mine."
A general practitioner, supportive of the machine, feared the tasks its masters at WellPoint would use it for: "If I could have Watson at my desk, I'd be happy. It takes a lot of time to look through info on eMedicine, UpToDate, etc., depending on how familiar I am with the information to start with. But putting it to use for an insurance company -- bad idea."
Some took it a step further, suggesting (with various degrees of irony) that Watson signals the beginning of the end for humans. Many invoked science fiction films such as 2001 and The Terminator to demonstrate the dangers of artificial intelligence.
A surgeon summed it up with a dire forecast: "When the computers have it all figured out and can reproduce then they won't need us."
Some physicians sat somewhere in the middle of this debate, recognizing both the danger and the rewards that Watson may bring. "It can be a diagnostic aid for physicians; unfortunately, my cynical side tells me it will be marketed as a replacement for the personal face-to-face visit between a physician and patient," hedged one surgeon.
The effectiveness and the consequences of Watson are likely to be debated and analyzed as they continue to develop. But the use of powerful information technology seems likely to become a permanent part of medical practice.
The full discussion of this topic is available at: http://boards.medscape.com/forums/.2a0d8d05/37. Note, this is open to physicians only.
Δεν υπάρχουν σχόλια:
Δημοσίευση σχολίου