FDA permits marketing of AI algorithm for detecting wrist fractures | Hacker News Reader

FDA permits marketing of AI algorithm for detecting wrist fractures

fda.gov

95 pointsgk118 comments

Interesting! I hope the tech works and can reduce potential suffering from mis-read x-rays.

My brother fractured his scaphoid bone during a high school lacrosse game, and the doctor who x-rayed his wrist did not detect it.

He continued living an active life with a broken wrist (he has a high pain tolerance) for several months before the worsening pain sent him to an orthopedic specialist.

Being unaware of the fracture, he had done quite a lot of damage to the bones in his wrist, which wound up needing multiple surgeries and nearly 18 months (!) in a cast to fully heal. He also permanenly lost a bit of that wrist’s range of motion.

I welcome any tool that helps doctors not miss this sort of thing!

Nice. I was going to post, 'how hard can it be to identify a broken wrist', but you answered me that it does happen.

Apparently it can be really difficult. I had a fall in my house onto an outstretched hand, my doc thought it was broken so she sent me for X-rays. She also told me that it is really easy to miss scaphoid fractures, so if it still hurts in two weeks to come back for more X-rays because sometimes the fracture can be easier to image as it heals. They also said that might not work and if I am still in pain without a diagnosed fracture at week 6 that I would need an MRI.

Thankfully it wasn't broken at all, it just hurt like hell for a week.

I had the almost opposite experience when I fractured a bone in my hand (it was a minor bone chip).

The tech at the immediate care facility saw it, but my ortho didn't when he reviewed it on a follow-up. I had to point out the bone and where it was before my ortho saw it. His response was basically "Oops" followed by removal of the temporary cast/sling and replacing all of that with a reinforced removable brace (which was awesome, because I could easily shower).

I broke my hand after dumping my bicycle taking a corner too fast on a training ride (I was doing around 25 mph) and losing traction under my rear tire. Rode home about 5 miles on that broken hand...

nothing is permanent

You may be technically correct, but what’s the point of this comment?

For as long as he is alive, he will have less range of motion in his right wrist than his left, barring advances in nanotechnology bone reconstruction or something.

Millions of years from now, when his body has broken down into petrochemicals, I suppose that yes, his wrist will then have a greater range of motion then it does now.

But that’s hardly relevant.

> For as long as he is alive, he will have less range of motion in his right wrist than his left, barring advances in nanotechnology bone reconstruction or something.

If that is his attitude - and the attitude of everyone around him - then that will likely be the case.

In 'reality' - whatever that means to you - plenty of folks have demonstrated that the body is malleable.

With work, ROM is restorable.

The word 'permanent' doesn't mean what you seem to think it means.

Does this mean that from now on, anyone wanting to file some flavour of AI classifier for medical diagnosis, can just file a 510k because there is a precedent?

Only if your intended use also remains the same (i.e. for detecting wrist fractures).

Some more detail (from https://www.fda.gov/MedicalDevices/DeviceRegulationandGuidan...): What is Substantial Equivalence

A 510(k) requires demonstration of substantial equivalence to another legally U.S. marketed device. Substantial equivalence means that the new device is at least as safe and effective as the predicate.

A device is substantially equivalent if, in comparison to a predicate it:

has the same intended use as the predicate; and has the same technological characteristics as the predicate; or has the same intended use as the predicate; and has different technological characteristics and does not raise different questions of safety and effectiveness; and the information submitted to FDA demonstrates that the device is at least as safe and effective as the legally marketed device. A claim of substantial equivalence does not mean the new and predicate devices must be identical. Substantial equivalence is established with respect to intended use, design, energy used or delivered, materials, chemical composition, manufacturing process, performance, safety, effectiveness, labeling, biocompatibility, standards, and other characteristics, as applicable.

No, it's contingent on what exactly the classifier does and how independent of third party clinical judgement it is. The more hands off the clinician is, the more dangerous a false negative is, and the more time sensitive the result is, the more likely your diagnostic is going to be class 3.

I would naively expect the bulk of new low risk ML product submissions to be de Novo submissions due to lack of real precedent.

As someone who has had to deal with a hard to read break in an x-ray this is great news.

Had the ER docs had this available I wouldn't have had to sit in agonizing pain and uncertainty for a day before getting into see a surgeon.

It wasn't wrist related, but it seems like the application of this to other stuff is not far off.

Very good news

Url changed from https://interestingengineering.com/fda-approves-ai-tool-that..., which unless I'm missing something is cribbed entirely from this press release.

And in another 5-10 years insurance might approve it. Until then it'll probably be a cash price of anywhere from $400-$4000 like the cost of an MRI.

This is a tool for use by doctors, it's not going to be billed for separately.

>And in another 5-10 years insurance might approve it

Not sure why you think that, but I work in a similar field and I promise you you're wrong.

Because we totally use MRIs to detect wrist fractures.

Reply on news.ycombinator.com