1

Topic: Machine intelligence VS doctors

On Tuesday carried out action. On platform Skychain developed by us 3 neural networks for diagnostics of a cancer of a breast (under the image ),  under the image with , heart diseases on an electrocardiogram have been trained. Everywhere  exceeded (though and it is not strong) doctors - narrow experts. Short video - https://www.youtube.com/watch?v=Iyd3ujEfTTQ still detailed Will be fast. I am surprised, there is a cancer of lungs of the first stage is passed by doctors on a X-ray in 70 % of cases!! It . The Same subject with many other illnesses. AI is capable to reduce an error in times. Especially in those places, where not enough the qualified doctors (for the Russian Federation it everywhere, except Moscow). How consider, soon AI replaces doctors?

2

Re: Machine intelligence VS doctors

S> I am surprised, there is a cancer of lungs of the first stage is passed by doctors on a X-ray in 70 % of cases!! It it is ripened. Like IBM Watson already showed that better average doctors. So never a surprise.

3

Re: Machine intelligence VS doctors

S> As consider, soon AI replaces doctors? Replaces - is not present, adds - yes.

4

Re: Machine intelligence VS doctors

Hello, ononim, you wrote: S>> I am surprised, there is a cancer of lungs of the first stage is passed by doctors on a X-ray in 70 % of cases!! It it is ripened. O> like IBM Watson already showed that better average doctors. So never a surprise. Here there were more precisely doctors of medical sciences. And it only a prototype! Watson assigns treatment, computer sight at it is not present. Plus   - 200-1000$. Plus for operation with it you should at  buy and deliver yourself pieces of iron for ten million $. In general  in my opinion

5

Re: Machine intelligence VS doctors

Hello, Muxa, you wrote: S>> As consider, soon AI replaces doctors? M> replaces - is not present, adds - yes. About adds - a question solved. We also make it. I here suspect 20 years forward. I think, whether can  replace? You come into a diagnostic capsule, you were studied at once by the robot, if needed treated, made a tightening, extended  , stuck vitamins of minerals, and go further for 5 years looking younger.

6

Re: Machine intelligence VS doctors

S> Watson assigns treatment, computer sight at it is not present. S> Plus   - 200-1000$. Plus for operation with it you should at  buy and deliver yourself pieces of iron for ten million $. In general  in my opinion Mda, and after all could become rich, exposing in personal use for diagnostics/treatment of small sores. Large too, but in the form of the recommendation to address to the doctor.

7

Re: Machine intelligence VS doctors

Hello, ononim, you wrote: S>> Watson assigns treatment, computer sight at it is not present. S>> Plus   - 200-1000$. Plus for operation with it you should at  buy and deliver yourself pieces of iron for ten million $. In general  in my opinion O> Mda, and after all could become rich, exposing in personal use for diagnostics/treatment of small sores. Large too, but in the form of the recommendation to address to the doctor. Greed and aiming on  .

8

Re: Machine intelligence VS doctors

Hello, sharpcoder, you wrote: S> As consider, soon AI replaces doctors? The doctor of sciences well noticed that the doctor on one picture does not accept the decision. And where you got access to images for training of your networks?

9

Re: Machine intelligence VS doctors

S> I here suspect 20 years forward. I think, whether can  replace? Is not present S> you come into a diagnostic capsule, you were studied at once by the robot, if needed treated, made a tightening, extended  , stuck vitamins of minerals, and go further for 5 years looking younger. It differs from " trained on recognition of tumors in a picture" a little

10

Re: Machine intelligence VS doctors

Hello, alsemm, you wrote: A> Hello, sharpcoder, you wrote: S>> As consider, soon AI replaces doctors? A> the doctor of sciences well noticed that the doctor on one picture does not accept the decision. A row of doctors to the patient do not adjoin. These are those who looks , , ,  and so forth Another  to me personally (not on video) told that considers as a principal problem of doctors that they select at once a direction and normally narrow down thinking already being guided by initiating judgement. A> and where you got access to images for training of your networks? Is generally available . Here at a forum prompted, I . A cancer of lungs could not train, thus that  was huge... But already there is an understanding as model to rebuild.

11

Re: Machine intelligence VS doctors

Hello, sharpcoder, you wrote: S> As consider, soon AI replaces doctors?

12

Re: Machine intelligence VS doctors

Hello, sharpcoder, you wrote: S> Another  to me personally (not on video) told that considers as a principal problem of doctors that they select at once a direction and normally narrow down thinking already being guided by initiating judgement. Such not only with doctors. S> a cancer of lungs could not train, thus that  was huge... But already there is an understanding as model to rebuild. The good project at you turns out. Good luck!

13

Re: Machine intelligence VS doctors

Hello, sharpcoder, you wrote: And what if you  directs the doctor on a false way and the person sets the fads? If for false information which can give  its developers do not answer with the head that such system it is impossible to implement.

14

Re: Machine intelligence VS doctors

Hello, sharpcoder, you wrote: S> I am surprised, there is a cancer of lungs of the first stage is passed by doctors on a X-ray in 70 % of cases!! It it is ripened. And how many false actuatings?

15

Re: Machine intelligence VS doctors

Hello, alsemm, you wrote: A> Hello, sharpcoder, you wrote: S>> the Cancer of lungs could not train, thus that  was huge... But already there is an understanding as model to rebuild. A> the good project at you turns out. Good luck! Thanks!!

16

Re: Machine intelligence VS doctors

Hello, turbocode, you wrote: T> Hello, sharpcoder, you wrote: T> And what if you  directs the doctor on a false way and the person sets the fads? T> if for false information which can give  its developers do not answer with the head that such system it is impossible to implement. You think, it is better let 70 die from for errors of the doctor, than 20 of for errors of the doctor armed ?

17

Re: Machine intelligence VS doctors

T>> And what if you  directs the doctor on a false way and the person sets the fads? T>> if for false information which can give  its developers do not answer with the head that such system it is impossible to implement. S> you think, it is better let 70 die from for errors of the doctor, than 20 of for errors of the doctor armed ? You still will need to prove it to Ministry of Health. Because it can appear that with  will to die even more. But an output I think is: if the patient signs a piece of paper that it not against that that it will be parallel to look also  and to produce recommendations to the doctor.

18

Re: Machine intelligence VS doctors

Hello, Real 3L0, you wrote: R3> Hello, sharpcoder, you wrote: S>> I am surprised, there is a cancer of lungs of the first stage is passed by doctors on a X-ray in 70 % of cases!! It it is ripened. R3> and how many false actuatings? At the left results of people (the first test): the Second tests (new triple of doctors):

19

Re: Machine intelligence VS doctors

Hello, turbocode, you wrote: T> Hello, sharpcoder, you wrote: T> And what if you  directs the doctor on a false way and the person sets the fads? And doctors never are mistaken? Or what is necessary to Jupiter, is not necessary to a bull? If the program is mistaken is is not admissible, and - it is possible for the person? In this case it is enough to receive system which statistically gives not less authentic forecast, than the average doctor. The standard dial-up of statistical metrics allows to estimate efficiency and applicability of system. T> if for false information which can give  its developers do not answer with the head that such system it is impossible to implement. It at all so. If the person worries over each action, the result does not turn out better. On the contrary, for many progressive decisions it is necessary to risk, it allows to receive more effective system in the future and on big enough segment the amount of errors will be less.

20

Re: Machine intelligence VS doctors

T>> And what if you  directs the doctor on a false way and the person sets the fads? LP> and doctors never are mistaken? Or what is necessary to Jupiter, is not necessary to a bull? If the program is mistaken is is not admissible, and - it is possible for the person? The doctor can be condemned, and the program is not present.

21

Re: Machine intelligence VS doctors

Hello, turbocode, you wrote: T>>> And what if you  directs the doctor on a false way and the person sets the fads? LP>> and doctors never are mistaken? Or what is necessary to Jupiter, is not necessary to a bull? If the program is mistaken is is not admissible, and - it is possible for the person? T> the Doctor can be condemned, and the program is not present. That fact that it is impossible to condemn the program, does not affect in any way its operation. The program or gives the exact diagnosis, or is mistaken. The doctor is tired, the program - is not present. Operation of the doctor depends on its mood, the program yields stable result. The fear of the doctor before responsibility only levels these negative moments which initially miss at the program.

22

Re: Machine intelligence VS doctors

LP> That fact that it is impossible to condemn the program, does not affect in any way its operation. The program or gives the exact diagnosis, or is mistaken. LP> the doctor is tired, the program - is not present. Operation of the doctor depends on its mood, the program yields stable result. The fear of the doctor before responsibility only levels these negative moments which initially miss at the program. I think that doctors begin  and will rely more and more on decisions  to the first . On court the doctor tells that relied on results . Total: the doctor plant,  forbid.

23

Re: Machine intelligence VS doctors

Hello, sharpcoder, you wrote: S> I am surprised, there is a cancer of lungs of the first stage is passed by doctors on a X-ray in 70 % of cases!! It it is ripened. I am surprised, why AI for years train to play to a pillbox, but do not use to applied medicine. It is plot or there is any latent problem? https://www.youtube.com/watch?v=92tn67YDXg0.

24

Re: Machine intelligence VS doctors

Hello, turbocode, you wrote: T> And what if you  directs the doctor on a false way and the person sets the fads? Then give to a heap we forbid medical reference manuals and Google. Suddenly the doctor not opens that page.

25

Re: Machine intelligence VS doctors

Hello, turbocode, you wrote: T> the Doctor can be condemned, and the program is not present. In case of a cancer of a breast I so suppose the program specifies in a stain in a picture. The stain is examined. Consider that it is such special magnifier for weak-sighted doctors. And then undertakes  - without it unless it is possible to begin treatment?? I am am surprised here with people who cannot distinguish a fur-tree from a fir. Probably, among them there are also oncologists, suppose - they generally bad doctors as pattern recognition does not work for them.