Eric Topol looks at a bunch of large trials (tens of thousands of patients) of AI/LLM in mammography and concludes all mammograms should be aided by these tools.
-
Eric Topol looks at a bunch of large trials (tens of thousands of patients) of AI/LLM in mammography and concludes all mammograms should be aided by these tools.
Important to note these studies extend back 6+ years in different countries & include gold-standard randomized prospective trials.
I think it's very easy to fall into the 'AI is always bad' thinking but the evidence in favor here looks strong.
https://erictopol.substack.com/p/why-all-mammograms-should-incorporate
-
T #medicine shared this topic
-
Eric Topol looks at a bunch of large trials (tens of thousands of patients) of AI/LLM in mammography and concludes all mammograms should be aided by these tools.
Important to note these studies extend back 6+ years in different countries & include gold-standard randomized prospective trials.
I think it's very easy to fall into the 'AI is always bad' thinking but the evidence in favor here looks strong.
https://erictopol.substack.com/p/why-all-mammograms-should-incorporate
@brianvastag This is the kind of use of AI that actually makes sense.
-
Eric Topol looks at a bunch of large trials (tens of thousands of patients) of AI/LLM in mammography and concludes all mammograms should be aided by these tools.
Important to note these studies extend back 6+ years in different countries & include gold-standard randomized prospective trials.
I think it's very easy to fall into the 'AI is always bad' thinking but the evidence in favor here looks strong.
https://erictopol.substack.com/p/why-all-mammograms-should-incorporate
@brianvastag I see it as a scientific tool, like a spectrometer. I have also seen a (very) few artists use it as a *medium*, like someone would use a stretched canvas or some film stock.
That's not the same as some CEO using it to "create" the company's new brochure, instead of hiring someone to do it.
-
@brianvastag This is the kind of use of AI that actually makes sense.
My wife had been involved in this side of AI for over a decade.
One axe I will consistently grind is that when talking about AI, we should be specific about what kind.
Much of the current backlash is against GenAI like LLMs, but there's a wide array of things that could be labelled as AI.
Sometimes it's as basic as linear regression, a technique that predates computers by a long shot.
-
My wife had been involved in this side of AI for over a decade.
One axe I will consistently grind is that when talking about AI, we should be specific about what kind.
Much of the current backlash is against GenAI like LLMs, but there's a wide array of things that could be labelled as AI.
Sometimes it's as basic as linear regression, a technique that predates computers by a long shot.
@dbattistella @hewer_of_code true...as scientific tools, lots of systems labeled AI have been and are being used...neural networks research dates back decades of course...
-
@dbattistella @hewer_of_code true...as scientific tools, lots of systems labeled AI have been and are being used...neural networks research dates back decades of course...
Yup. Already in the mid 90s, I had the option to study neural networks (which I didn't end up doing).
