readee456 · HackerTrans

vor 5 Jahrendiscuss

I can't help but think some of the barriers here involved proving the software in a situation decidedly different than a clinical setting. I would not be surprised if an immigration medical officer developed different views about diseases than a GP or ER doctor. They're not treating the person, they're not in a doctor-patient relationship with the person, they're not really even "diagnosing" the person, they're just deciding whether they're "too sick" to come into the country. Maybe if the person looks messed up in some other way, their chest x-ray gets interpreted a little more strictly.

readee456

vor 5 Jahrendiscuss

I'm pretty sure people said the same things (nothing will ever change, doctors will never advocate for or accept change) when radiology went from films to digital. I'm sure they said the same things when radiology went from having scribes to using voice recognition software (e.g. Nuance) for reports.

There seems to be a misconception that this is some kind of "all or nothing" thing, where AI will "automate away" radiologists. It's like a decade ago when everybody thought we were just about to "automate away" human drivers, except unlike driving, most radiology reads are by definition (i.e. a sick person) exceptional, out-of-baseline scenarios.

I think this is missing some things about radiology economics. There are indeed incentives to automate as much as possible, especially for outsourced radiology practices like Radiology Partners or people getting paid by pharma companies for detailed clinical trial reads. Organizations like these are getting paid a certain amount per read. If they can use software to speed up parts of their work while demonstrating effectiveness, they make more money. Eventually this drives down the price. There would still be a human in the loop to review or sign off on whatever the AI does, and to look for any anomalies that it misses. But there can be less time spent on rote work or routine segmentation, and more on the overall disease picture.

It's true the amount of imaging going on in the US has increased faster than both the population growth and the number of radiologists. At a certain point, the number of existing radiologists doesn't have time to read the images at any price. This gives the alleged cartel a few choices: graduate more radiologists, outsource reads, or use software to produce more output per radiologist. In the last case, which a self-interested group would obviously choose, they get paid the same but each individual patient pays less.

comments