News for hackers. Stuff that matters.

A comment on Use of artificial intelligence for image analysis in breast cancer screening. Parent comment

Re:Use of artificial intelligence for image analysis in brea... by somenewaccount1 on Thursday September 02, 2021 @02:47PM (#28395812)
0% of the studies were better than 2 radiologists reviews - which is the current practice.

The 6% only scored better when compared to a single radiologist review. That's why we use double reviews.

[ Parent | on HN ]
Re:Use of artificial intelligence for image analysis in brea... by TameAntelope on Thursday September 02, 2021 @03:31PM (#28396308)
What about 1 AI and 1 radiologist, then? Double the capacity for reviews.
[ Parent | on HN ]
Re:Use of artificial intelligence for image analysis in brea... by kenjackson on Thursday September 02, 2021 @04:23PM (#28396887)
Or the consensus of two AIs.
[ Parent | on HN ]
Re:Use of artificial intelligence for image analysis in brea... by sciclaw on Thursday September 02, 2021 @07:19PM (#28398572)
Yes, but when disagreements occur, it's hard to make AI systems "talk" and convince one another
[ Parent | on HN ]
Re:Use of artificial intelligence for image analysis in brea... by lavishlatern on Thursday September 02, 2021 @04:43PM (#28397105)
> That's why we use double reviews.

We being Europeans. Double reads are not standard in the majority of the world, including the US.

[ Parent | on HN ]