Reading CT scans is incredibly hard. I have no doubt that with sufficient training data, deep learning will eventually be better at it than trained specialists.
Show this thread
That is, until the model encounters something it hasn't quite seen before. Hence the need to keep humans in the loop...
-
-
Unless we start framing these problems as "open set" learning problems:https://arxiv.org/abs/1910.02830
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
You are back!
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
That is until an AGI with an equivalent life of experience as many humans can again outperform a person on that edge case. Many many years away.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
But to be effective humans need sophisticated repeated training too. Which is especially hard when most of their work will be taken out by some model. Oups...
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
For MRI etc we need some available, standardized datasets sponsored by a big company. Files are very large, different scanners have different formats, healthcare privacy issues etc.
-
I'm sure this is going on behind closed doors. But providing researchers with the data and compute power needed could dramatically change healthcare in 10-20 years.
End of conversation
New conversation -
-
-
imagine a 1000-slice scan of the abdomen, for example. Do we really expect the doctor would carefully examine every little corner of every slice? I can envision the following scenario as being the future of AI-diagonistics: a doctor looks at the scan, makes a tentative diagnosis,
-
and then they'd look at all the diagnostic-alarms generated by an AI system. Normally, these would be either false-alarms or pathological tissue already identified by the doctor. Occasionally, however, the doctor would say to themselves: "how did I miss that spot?!".
- Show replies
New conversation -
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.