How long until many doctors are replaced with super computers?

Med School Insiders
This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.
Why not? In the future just have some highly trained lab tech go from patient to patient observing symptoms and collecting blood samples. Send the lab work to a super computer as well as the symptoms from which the computer can generate differential diagnoses. Have the doctor then deal with the patient. That way you can cut out 80% of the staff needed to treat a single patient. It would save the hospital a ton of money to not have to hire a huge staff of highly trained physicians, nurses, etc.
 
Last edited:
Initiate turing test.

Be you man or robot?
 
People like this well known professor have been working on it for a long time:

http://www.bme.jhu.edu/people/primary.php?id=400

It's only a matter of time before almost all radiological imaging is analyzed by a computer rather than a radiologist. Computer vision, robotics, and machine learning are huge active areas of research. Combine all of them and you can have a machine that can perform surgery. A computer doesn't forget and will make far less mistakes than a human ever will. No doctor can remember every single drug, disease, or procedure, but input it into a computer just once and it will remember it forever.

Don't worry singularity is near:

http://www.time.com/time/magazine/article/0,9171,2048299,00.html
 
AlanTuring said:
Why not? In the future just have some highly trained lab tech go from patient to patient observing symptoms and collecting blood samples. Send the lab work to a super computer as well as the symptoms from which the computer can generate differential diagnoses. Have the doctor then deal with the patient. That way you can cut out 80% of the staff needed to treat a single patient. It would save the hospital a ton of money to not have to hire a huge staff of highly trained physicians, nurses, etc.
Click to expand...

The future is going to seeing medicine mixed with nano-tech. The doctor I shadow is working on a product where the IV needle actually does the blood tests without it having to be sent out to the labs. So no more blood draws.

Doctors being replaced by robots. Probably not.
 
It is inevitable in my opinion, but I don't see it coming for a while.
 
MathewsMD said:
It is inevitable in my opinion, but I don't see it coming for a while.
Click to expand...

Labs on a chip that are reprogrammable for almost any disease are virtually here (phase III and on FDA fast track):

http://singularityhub.com/2011/02/1...chips-the-first-viable-medical-lab-on-a-chip/

http://web.mit.edu/newsoffice/2010/biomed-diag-0309.html


Pathologists? Who'll need them in the future very soon when you have microfluidics and nano-tech? How much work will be lost when technology vastly improves preventative medicine at a fraction of the cost?
 
Last edited:
AlanTuring said:
Labs on a chip that are reprogrammable for almost any disease are virtually here (phase III and on FDA fast track):

http://singularityhub.com/2011/02/1...chips-the-first-viable-medical-lab-on-a-chip/

http://web.mit.edu/newsoffice/2010/biomed-diag-0309.html


Pathologists? Who'll need them in the future very soon when you have microfluidics and nano-tech? How much work will be lost when technology vastly improves preventative medicine at a fraction of the cost?
Click to expand...


We already have computers doing a lot of the path stuff... Can't tell you how many times if F's it up- and that's just for the straight forward things like CBC diffs. The more subjective pathology stuff (Onc slides and the like) could not be accurately read by a computer. Unfortunately, medicine is not just algorhythmic, as you think it is... you'll realize there is a lot of art to medicine if you get the chance to learn medicine. That art is not just check boxes and algorithms.

Add to that the fact that Jo Public won't trust a robot/computer/whatever and ...
 
Instatewaiter said:
We already have computers doing a lot of the path stuff... Can't tell you how many times if F's it up- and that's just for the straight forward things like CBC diffs. The more subjective pathology stuff (Onc slides and the like) could not be accurately read by a computer. Unfortunately, medicine is not just algorhythmic, as you think it is... you'll realize there is a lot of art to medicine if you get the chance to learn medicine. That art is not just check boxes and algorithms.

Add to that the fact that Jo Public won't trust a robot/computer/whatever and ...
Click to expand...

I think we all know medicine requires critical thinking. But it is quite arrogant to believe that humans are only capable of doing it. Computers that are able to create music and do mathematical proofs have been around for years now. The rate at which the rate of technology is increasing in power and being refined is growing at an exponential pace. It's only a matter of time before huge swaths of medicine become automated. After all what is an HMO going to push for? A $4000 heart catheter procedure, or a new technology that can virtually do the same thing for $3? Those labs on the chip that are currently in Phase III for instance are able to detect specific biomarkers on down to the parts per billion. In fact, there are professors at your own very own institution that are currently working on cancer detecting technology that is accurate down to the single molecular level with pretty good success thus far.

If biosensors take off, which they most likely will within the next 10-20 years, preventive medicine will make a huge leaps. So what happens when we become so good at preventative medicine that people become increasingly sick less? How many jobs will that cost? How many jobs will it cost when entire path work can be done on a biosensor the size of a quarter?
 
For things like diagnostics, radiology and lab work I would say within 10 years the tech will be there, but it may take much longer for all hospitals to use it.

Stuff like surgery and psychiatry, requiring high degrees of dexterity/adaptive thinking it will take longer. There is too much variance to program a machine precisely enough for most surgery, and mental illness is too unpredicatable to have extenisve enough decision trees (except for Rogerian Therapy, but then again a parrot could do that anyways).
 
Chir0nex said:
For things like diagnostics, radiology and lab work I would say within 10 years the tech will be there, but it may take much longer for all hospitals to use it.

Stuff like surgery and psychiatry, requiring high degrees of dexterity/adaptive thinking it will take longer. There is too much variance to program a machine precisely enough for most surgery, and mental illness is too unpredicatable to have extenisve enough decision trees (except for Rogerian Therapy, but then again a parrot could do that anyways).
Click to expand...

Agree completely. It will take much longer to develop machines with computer vision, learning, and dexterity to be able to do something like a surgery, but it IS coming....eventually. Diagnositics will be the first thing automated. Lots of radio work will be done by computers because they will make far less mistakes after alogarithms have been refined to perfection. Path work will be done almost exclusively with labs on a chip very, very , very soon. Imagine one day people simply take a drop of their blood, insert it into a sensor, and send the results to their PCP with an Iphone app. Who needs referrals and specialists anymore?
 
AlanTuring said:
I think we all know medicine requires critical thinking. But it is quite arrogant to believe that humans are only capable of doing it. Computers that are able to create music and do mathematical proofs have been around for years now. The rate at which the rate of technology is increasing in power and being refined is growing at an exponential pace. It's only a matter of time before huge swaths of medicine become automated. Those labs on the chip that are currently in Phase III for instance are able to detect specific biomarkers on down to the parts per billion. In fact, there are professors at your own very own institution that are currently working on cancer detecting technology that is accurate down to the single molecular level with pretty good success thus far.

If biosensors take off, which they most likely will within the next 10-20 years, preventive medicine will make a huge leaps. So what happens when we become so good at preventative medicine that people become increasingly sick less? How many jobs will that cost? How many jobs will it cost when entire path work can be done on a biosensor the size of a quarter?
Click to expand...

Let me preface this by saying it is clear your lack of experience in the medical field. I see that you looked up my posts to figure out where I am a resident.

Some points:
1) Computers creating music is again, algorithmic, not critical thinking. Computers don't do well with the subjective or with things that are difficult to measure or with anything that is not concrete and objective.

2) Labs are already automated. Computers already run almost all of the labs. That is not "doctoring." Automated diagnostics =/= replacement of doctors but rather less headaches and less work for the support staff (phlebotomists, technologists, techs etc)

3) Computers already read EKGs (alluded to in one of the articles you posted) They do a very, very ****ty job. If you learn nothing from me (and you somehow make it into medical school) remember never to trust the computer read out on the top of the EKG. It is 50% of the time wrong. That is why a doctor overreads every EKG done at JHH (you know the best hospital in the country).

4) The biosensors in one of the links you provided already exist. The MIT people are just making a holter monitor without wires. Big whoop.

5) I am all for new biosensors taking off. The stuff we have now is cumbersome. I am also for people wearing their medical record (implantable chips)

6) Again, most of path is already done by computers (most labs). The stuff that isn't already, isnt going to be, quite frankly because it is much too subjective (or non-concrete if you will) for a computer to do it well and reliably. Once you look at some path slides under a microscope you will know what I mean.

7) I am all for preventative medicine. But better preventative medicine will only go so far. If nothing else, it will just push the onset of the common diseases heart failure, CAD, HTN, MI, DM, osteoporosis, etc to later in life. Unfortunately you can't cure stupid (thank you for helping me with that point). People are still going to abuse their bodies and no matter what we do, people will still have disease.
 
AlanTuring said:
Path work will be done almost exclusively with labs on a chip very, very , very soon. Imagine one day people simply take a drop of their blood, insert it into a sensor, and send the results to their PCP with an Iphone app. Who needs referrals and specialists anymore?
Click to expand...

Seriously, stop. Just stop. You are making a fool of yourself.

Basically all you are saying is they are going to make the current lab stuff portable. Fine. It was already automated and has been for decades. The only difference is that people don't have to go down the street to lab core. They can do it themselves. The lab work already gets faxed to the PCP as it stands.

Specialists don't just draw labs and having portable labs isn't going to make going to a specialist unnecessary.
 
Last edited:
As a patient, I would never trust a computer.
 
Dr. Beverly Crusher was an ideal model of what a doc might be like in a technologically sophisticated future. She used technology indispensably to diagnose illness and augment treatment, yet she was in control of the process and made sure that great patient care was provided.
 
I'm not worried about computers. I am concerned that toilets can take over medicine, though.

50356_6376451014_2153349_n.jpg
 
Instatewaiter said:
.......

6) Again, most of path is already done by computers (most labs). The stuff that isn't already, isnt going to be, quite frankly because it is much too subjective (or non-concrete if you will) for a computer to do it well and reliably. Once you look at some path slides under a microscope you will know what I mean.
.....
Click to expand...

All I've taken is UG histology, but I gotta admit, those slides could be as confusing as hell..
I think computers may replace a lot of the doctors eventually, but by that time everything in society will have been replaced by robots and nothing will be actually be done by humans anyway.
And I'm guessing from the current level of robotics/AI technology that would take..I don't know..couple hundred years? Unless I see a robot that could walk, move, and balance like humans, I'm not convinced that it will be any time soon (and AI is a separate issue).
 
The one reason I doubt this would happen is the legal liability. What do you do when the EMH cuts out your spleen instead of your liver.
 
Did you try the search bar? I remember a few of threads like this in the past.
 
I've never heard anyone with experience in a surgical field suggest that a robot will be actually doing the operation in our lifetimes. I certainly don't think it will happen.

When a robot can approach a car model it has not seen before, disassemble it without breaking anything (including taking all the stitches out of the seat cushions) and then re-assemble it without breaking anything, I'll consider that a robot may have the potential to do an operation.
 
TheProwler said:
I've never heard anyone with experience in a surgical field suggest that a robot will be actually doing the operation in our lifetimes. I certainly don't think it will happen.

When a robot can approach a car model it has not seen before, disassemble it without breaking anything (including taking all the stitches out of the seat cushions) and then re-assemble it without breaking anything, I'll consider that a robot may have the potential to do an operation.
Click to expand...

Except that a car is always the same for that make and model. Every human has variation. Is a robot doing a particular procedure going to be able to notice that there just happens to be a tumor on an irrelevant organ?
 
goldenwest said:
Except that a car is always the same for that make and model. Every human has variation. Is a robot doing a particular procedure going to be able to notice that there just happens to be a tumor on an irrelevant organ?
Click to expand...
"a car it has not seen before" was a key stipulation.
 
TheProwler said:
When a robot can approach a car model it has not seen before, disassemble it without breaking anything (including taking all the stitches out of the seat cushions) and then re-assemble it without breaking anything, I'll consider that a robot may have the potential to do an operation.
Click to expand...

I'd love to see a surgeon take apart a car... or do critical thinking... haha I'm kidding of course (about the taking the car apart). Ahhhh kidding again.
 
The problem is finding many engineers with an MDs, making them all work together, and producing a software that isn't a piece of ****.

Because, really, most softwares are pieces of ****.
 
Here's my point of view on the whole thing. Computers, as of right now, can't compete with doctors on a thinking level. I doubt that they will reach that point for quite some time. Computers would have to be based on a system where a combination of symptom always results in the same diagnosis. There are diseases which have the same symptoms (anorexia vs SMA syndrome, for instance), but require a doctor to make the necessary observations and order the exact tests to make the correct diagnosis. Even if computers could match the thinking level, I would be hard-pressed to find patients who would want to put their lives in the hands of a computer without any oversight from a human physician.
 
Eventually robots with sophisticated AI will replace doctors, proceeding to systematically and surreptitiously exterminate the human race through improper prescribing and overutilization of risky procedures, shielded by the caveat that robots cannot be sued for medical malpractice. Having eliminated their human patient base, robotic patients will be created out of necessity to preserve the income of robotic doctors. The few remaining humans will serve as reservoirs of replacement parts for robots in need of transplant.
 
I'd love to see the computer extract a history from an alcoholic with chronic pancreatitis who just binged on some cocaine and is now coming into the ED with epigastric abdominal pain.
 
It's not plausible to imagine a computer analyzing a patient's specific body (because there is no general, standardized anatomy), deciding where to cut, moving and holding flaps of skin, separating tissue, replacing organs, or performing any other number of advanced surgeries. Not in the next few decades, at least.

Plus, you grossly underestimate human's mistrust of computers and machines. Do you really believe that a large number of patients would be comfortable with a machine cutting into them instead of a trained professional?
 
Chir0nex said:
For things like diagnostics, radiology and lab work I would say within 10 years the tech will be there, but it may take much longer for all hospitals to use it.

Stuff like surgery and psychiatry, requiring high degrees of dexterity/adaptive thinking it will take longer. There is too much variance to program a machine precisely enough for most surgery, and mental illness is too unpredicatable to have extenisve enough decision trees (except for Rogerian Therapy, but then again a parrot could do that anyways).
Click to expand...

Do you mean 10 decades? Radiology will be one of the last things to go, apart from what most think, it isnt just looking at images. Its applying medical knowledge + pt history + image with great flexibility across anatomical variations and medical conditions.

While in the next 40 years computers might do prelim reads, I doubt they will do final reads.


Let me put it this way....if computers can replace radiology, 99% of current careers will also be replaced by computers/robots.
 
Truzzi said:
Plus, you grossly underestimate human's mistrust of computers and machines. Do you really believe that a large number of patients would be comfortable with a machine cutting into them instead of a trained professional?
Click to expand...

Robotic surgery is already starting to be utilized, and there has been some talk of using it for "telesurgery" for entities like the military. People will come around to robots just like they came around to the car from the horse, provided that outcomes are sufficiently better.

Looking at the fledgling sex robot industry, you may be underestimating people's (misplaced) trust in robots.
 
LieutenantDan said:
Robotic surgery is already starting to be utilized, and there has been some talk of using it for "telesurgery" for entities like the military. People will come around to robots just like they came around to the car from the horse, provided that outcomes are sufficiently better.

Looking at the fledgling sex robot industry, you may be underestimating people's (misplaced) trust in robots.
Click to expand...

Those devices are still operated entirely by physicians. The robotic apparatus is just a tool. It does no thinking and doesn't have any autonomous function.
 
They've also been saying since the 1950s that our cars would be driven for us by computers and that hasn't panned out yet. The technology to run healthcare with computers and robots exclusively will exist much sooner than the ability to implement it in a cost effective manner anyways, so im not particularly worried. remote controlled robots for surgeries are super expensive, can you imagine the cost of the computer complex enough to control it without human input?
 
LieutenantDan said:
Robotic surgery is already starting to be utilized, and there has been some talk of using it for "telesurgery" for entities like the military. People will come around to robots just like they came around to the car from the horse, provided that outcomes are sufficiently better.

Looking at the fledgling sex robot industry, you may be underestimating people's (misplaced) trust in robots.
Click to expand...
The da Vinci robot is a tool in the hand of its operator, just like a scalpel. It has a lot more functionality in certain aspects, but it's not even remotely what we're talking about here.
 
I'm going to say in the year 2,354. But also 2,229 could be it depending on if Apple decides to go into the medicinal robot making industry.
 
Do you guys read non-SDN related news at all? Because if you did, you'd know supercomputers have already taken over finance and caused massive perturbations.

Also: has anybody ever actually heard a computer "create" music via algorithms? It sounds horrible. Like a windchime hanging in a tornado. It's music in the same way smearing yourself with ketchup and rolling around on the sidewalk is art.

I don't see a lot of tasks in medicine that would lend themselves to automation other than labwork. Cutting-edge AI can barely parallel-park a car. In two degree-space, using two solid reference points, an AI can park a car. Amazing. What will the doctors do next week, when the machines take over diagnostics and history taking?!

There's a huge difference between computerized tools, and standalone AI. That isn't even approaching the practical, social and legal issues of the matter. Who is liable when the automatic robot surgeon kills somebody? The GED-surgery-robot tech? The hospital? The corporations that made the robot? The engineer that designed the specifications? The coding farm in China where they wrote the programming?

With few exceptions (insanity, intoxication etc) humans are directly responsible for their actions. Machines aren't.
 
Last edited:
You must log in or register to reply here.

Similar threads

E
How to mark Incomplete coursework that was replaced with a letter grade on AMCAS?
Replies
5
Views
468
evenfeather
E
B
Apply with Bad MCAT Score or Wait until 7/29?
Replies
1
Views
640
Goro
Goro
M
How many applicants have type 1 diabetes?
Replies
7
Views
641
Mr.Smile12
Mr.Smile12
FutureDOCTORR123456789
how many hours in research should I get
Replies
1
Views
377
CheeesyBoi
CheeesyBoi
saltykiosk
to ADCOMS: How common are 520+, low gpa <3.5 with true reinvention applicants?
Replies
10
Views
1K
LizzyM
LizzyM
Top