IBM’s Watson Could Diagnose Cancer Better Than Doctors

[Brainwrong]

IBMs Watson Could Diagnose Cancer Better Than Doctors | Republic of Innovation

As of now, Watson has assimilated over 600,000 unique types of medical evidence. In addition, Watson’s database includes two million pages sourced from a variety of different medical journals. To improve the link between symptoms and a diagnosis, Watson also has the ability to search through 1.5 million patient records to learn from previous diagnoses. This amount of information is more than any human physician can learn in a lifetime.

Pretty fucking cool!


Sent from my iPhone using Tapatalk

 

[egb_hibs]

This is right in AIs sweet spot. - pattern matching - and not really a surprise. Diagnostic medicine is one of those professions (or elements of a profession) that will be replaced by technology before long. Middle class jobs will be falling like flies over the coming decades - even more than less well off ones.

 

[Brainwrong]

I reckon this liberates doctors to do the human side, take more time, be afforded the luxury of really caring for their patients.

Not just removed from the workforce by binary baddies.


Sent from my iPhone using Tapatalk

 

[egb_hibs]

I reckon this liberates doctors to do the human side, take more time, be afforded the luxury of really caring for their patients.

Not just removed from the workforce by binary baddies.


Sent from my iPhone using Tapatalk

That's called nursing though.

Not that there is anything wrong with nursing - but this is the pattern we're likely to see; high skill and knowledge based jobs going, and softer skill ones hanging on a bit better. Probably one reason men are faring worse than women in some of these trends.

 

[Brainwrong]

That's called nursing though.

Not that there is anything wrong with nursing - but this is the pattern we're likely to see; high skill and knowledge based jobs going, and softer skill ones hanging on a bit better. Probably one reason men are faring worse than women in some of these trends.

Totally don't agree. Nursing requires soft skills that plenty of nurses are crap at, doesn't mean they can't do their jobs.

Doctors require soft skills that plenty of doctors are crap at. They too survive.

Both groups, including ones that ARE really good at their soft skill, are not afforded the leeway to spend time giving that care. It's mental care just as much as it is what you're talking about, the day to day physical care provided by a nurse.


Sent from my iPhone using Tapatalk

 

[egb_hibs]

Totally don't agree. Nursing requires soft skills that plenty of nurses are crap at, doesn't mean they can't do their jobs.

Doctors require soft skills that plenty of doctors are crap at. They too survive.

Both groups, including ones that ARE really good at their soft skill, are not afforded the leeway to spend time giving that care. It's mental care just as much as it is what you're talking about, the day to day physical care provided by a nurse.


Sent from my iPhone using Tapatalk

Not being funny then but could you describe what you envisage as the freed up caring doctor of the future, then? Everyone likes a doc with a good bedside manner but that is rather secondary to their main role, and is no more than nursing if the primary role is gone.

Broadly speaking primary roles are diagnostic and surgery - with general and specialist doctoring roles being either broad and shallow (Eg GP) or narrower and deeper examples of diagnostics (eg consultants) in most cases. If Watson becomes better at diagnostics then it surely will do at what most specialists do - which is diagnose within areas of expertise, whether cancer, bones, organs whatever.

Surgery is not a game for the touchy feely sorts and indeed is one or the careers where high quotas of socially functioning psychopaths are found.

You've obviously got less technical (than surgery) treatment expertise as well but again these skills are shared by nurses and paramedics, with the doctor's diagnostic expertise elevating his / her role - ie deciding which and how much drug to pump into a dying person rather than the technical skill of infecting that straight into a heart or whatever.

I'm honestly not sure what role you envisage doctors being freed up to do, which isn't actually a much reduced role closer to nursing or paramedic.

 

[Brainwrong]

Not being funny then but could you describe what you envisage as the freed up caring doctor of the future, then? Everyone likes a doc with a good bedside manner but that is rather secondary to their main role, and is no more than nursing if the primary role is gone.

Broadly speaking primary roles are diagnostic and surgery - with general and specialist doctoring roles being either broad and shallow (Eg GP) or narrower and deeper examples of diagnostics (eg consultants) in most cases. If Watson becomes better at diagnostics then it surely will do at what most specialists do - which is diagnose within areas of expertise, whether cancer, bones, organs whatever.

Surgery is not a game for the touchy feely sorts and indeed is one or the careers where high quotas of socially functioning psychopaths are found.

You've obviously got less technical (than surgery) treatment expertise as well but again these skills are shared by nurses and paramedics, with the doctor's diagnostic expertise elevating his / her role - ie deciding which and how much drug to pump into a dying person rather than the technical skill of infecting that straight into a heart or whatever.

I'm honestly not sure what role you envisage doctors being freed up to do, which isn't actually a much reduced role closer to nursing or paramedic.

No arguments on the surgery front.

Have you every met a doctor or heard of doctors saying 'Yep, plenty time to provide the care my patients need, totally not under any pressure. I'm totally cool with folk cancelling appointments too, plenty hours in the day to absorb that little inconvenience, I'm yet to feel like my patients are on a conveyor belt, escaping my grasp before I can fully diagnose their ills, I've gotta be all things to all men and I'm totally achieving that in every appointment'.

I'm saying things will actually improve because diagnosis will not cease to be required but will be positively augmented by the increased time afforded by things like Watson and doctors will be able to find shit a computer yet can't.

 

[egb_hibs]

No arguments on the surgery front.

Have you every met a doctor or heard of doctors saying 'Yep, plenty time to provide the care my patients need, totally not under any pressure. I'm totally cool with folk cancelling appointments too, plenty hours in the day to absorb that little inconvenience, I'm yet to feel like my patients are on a conveyor belt, escaping my grasp before I can fully diagnose their ills, I've gotta be all things to all men and I'm totally achieving that in every appointment'.

yes but the care they are talking about is the activity that is being replaced. What is the continuing activity once the technology matures?

I'm saying things will actually improve because diagnosis will not cease to be required but will be positively augmented by the increased time afforded by things like Watson and doctors will be able to find $#@! a computer yet can't.

this tech is in its infancy - it is one area where it is easy to imagine human input being completely redundant before long. Diagnostics is as I said above, right in the sweet spot. I don't expect computers to be writing good poetry anytime soon, but diagnostics medicine? Replaced.

 

[Brainwrong]

yes but the care they are talking about is the activity that is being replaced. What is the continuing activity once the technology matures?

Vastly improved doctory! There are real life nuances that as yet machines can't do and even experienced physicians miss. I'm talking about a holistic care system, roles may morph but the fundamentals remain.

this tech is in its infancy - it is one area where it is easy to imagine human input being completely redundant before long. Diagnostics is as I said above, right in the sweet spot. I don't expect computers to be writing good poetry anytime soon, but diagnostics medicine? Replaced.

I don't think jobs will completely remain as they are, because, otherwise what's the point? Frontiers that we haven't hit yet require humans at the forefront. So far, anyway. It's this freeing from laborious & technical diagnosis tasks (often resulting in errors) that I see as a major bonus to the accuracy of Watson.

 

[egb_hibs]

Vastly improved doctory! There are real life nuances that as yet machines can't do and even experienced physicians miss. I'm talking about a holistic care system, roles may morph but the fundamentals remain.

i feel like I'm being a dog with a bone, but so apologies, but again I don't know what this actually means. There will certainly be a need for continuing human input to research, but the number of practioners roles required will reduce massively. Machines are far less likely to miss things in diagnostics than even the most experienced human physician - once the technology has matured. What is an experienced doc doing? Pattern matching against learning and years of personal experience, and making inferences from all that. Computers can do this much, much more comprehensively.

I don't think jobs will completely remain as they are, because, otherwise what's the point? Frontiers that we haven't hit yet require humans at the forefront. So far, anyway. It's this freeing from laborious & technical diagnosis tasks (often resulting in errors) that I see as a major bonus to the accuracy of Watson.

what are the tasks that are not diagnostic? That's most of the job for most doctors! The labourious bit that's going away is the practice of medicine in most of its forms! As above, there'll still be a need for researchers working with the computer scientists, but much of the rest will become paramedical / nursing, I reckon?

 

[Brainwrong]

i feel like I'm being a dog with a bone, but so apologies, but again I don't know what this actually means. There will certainly be a need for continuing human input to research, but the number of practioners roles required will reduce massively. Machines are far less likely to miss things in diagnostics than even the most experienced human physician - once the technology has matured. What is an experienced doc doing? Pattern matching against learning and years of personal experience, and making inferences from all that. Computers can do this much, much more comprehensively.

what are the tasks that are not diagnostic? That's most of the job for most doctors! The labourious bit that's going away is the practice of medicine in most of its forms! As above, there'll still be a need for researchers working with the computer scientists, but much of the rest will become paramedical / nursing, I reckon?

I'm no doctor, but, I've met a few and knew a few trainees. I really didn't want to make this anecdotal, hence my resisting examples because, I don't work in this field, I can't speak on behalf of an entire profession. Although, I made an attempt with the 'struggles' side of things.

Face to face, body language, verbal history, language, none of these are interpretable by Watson. That's what I'm getting at. These things give insights comparisons with databases don't. You can't remove this aspect of diagnosis. Yet. And, I doubt for a very long time.

 

[Jack]

So who is this replacing? At least in this country?

The GP who makes the initial diagnosis then refers to the experts?

Nope, because folk probably won't know or even think they've got cancer when they rock up to the surgery.

The specialist in the hospital?

Nope, because even if this was the font of all cancer knowledge they'd still be needed for the treatments.

I can see it possibly being a supplement to the care pathway, it might be able to be more precise and therefore make treatment more efficient.

To be fair the experience I have of relatives and friends the NHS is frighteningly quick with cancer patients.

Is another step necessary? I wonder what the figures are for misdiagnosis by the GP.

Watson might need to go back to university and do the full medical course before its useful in real life.

 

[egb_hibs]

I've spoken with people involved in primitive efforts in this area years back. Even then computers were useful at diagnosing things gps wouldn't normally encounter - eg tropical diseases - you key in the symptoms and context (eg recent travel) and computers present potential diagnosis. Watson - which I have encounter in other fields - when it matures (which is always 'in the near future) could take this much further. I wouldn't bet big money on Watson itself getting there, but someone will - I'd think a google effort more likely, or someone like that. At that point it's straightforward versus other problems AI is touted as an answer to, to imagine computer guided questioning and keyed data matching diagnoses across vast data sets - as well as spotting transmission vectors etc. So yes, you'll need someone to man the machine but I wonder how much the doctors rule will be diluted.