https://jamanetwork.com/journals/jam...rticle/2814246

It has already happened. Medical doctors spend by far much more time playing with the software on the computers on their desks than they do interacting with their patients. Furthermore, rather than diagnosing, the younger ones are pushed to play the odds regarding problems based on the statistical probability of a diagnosis in the general population instead of analyzing the constellation of symptoms in the patient in front of them.

In the 1980s, there used to be decision trees called expert systems which could amplify medical doctors' skill and ability in regards to effecting a diagnosis based on observed symptoms. But that was not AI and is nothing more than a codification of the decision process which a skilled, experienced medical doctor would be going through anyway base on observations. However, it is much better at catching edge cases. It still requires though from the medical doctor and, especially, the ability to identify symptoms. Good luck in getting "AI" to differentiate between an abrasion and a rash etc, for example.

Now that I made the mistake of commenting, especially since the JAMA link seems dead, I belatedly ask what was the tie-in to Linux or even to FOSS?

Please don't "over-estimate" AI. It's just the latest "brand-new thing" in an industry which has seen a never-ending list of them.

The technology certainly has merit. But just remember: "There is no 'I' in 'AI!'"

There can be no doubt that in the early years of transition that there will occur LOTS of errors but hopefully machine learning can cause rapid updates that will soon eclipse considerable of the common human errors in AI use, including in medicine. Those early years will likely suffer from the fact that the initial programming is done by humans, programmer humans who may not have top notch skills in medicine and health care.

Attached Thumbnails

 

I agree. When a doctor engages in a competent practice of medicine they need to engage and assess the patient. The doctor looks for physical clues, behavioral clues, interprets laboratory tests as an underlying explanation for observed symptoms.
For example if the doctor notices someone who is severely withdrawn they would try to follow-up and find out why( depression, trauma, sleep deprived, etc.). AI is not capable of being able to diagnose with the level of sophistication that a human doctor can.

It just worked for me.

I posted in 'General'. There'll be FOSS AI eventually.

You have been absorbed.