How A US Hospital is Using AI to Analyze X-Rays - With Help From Red Hat

This week Red Hat announced one of America's leading pediatric hospitals is using AI to analyze X-rays, "to improve image quality and the speed and accuracy of image interpretation."

Red Hat's CTO said the move exemplifies "the positive impact AI can have in the healthcare field". Before Boston Children's Hospital began piloting AI in radiology, quantitative measurements had to be done manually, which was a time-consuming task. Other, more complex image analyses were performed completely offline and outside of the clinical workflow. In a field where time is of the essence, the hospital is piloting Red Hat OpenShift via the ChRIS Research Integration Service, a web-based medical image platform. The AI application running in ChRIS on the Red Hat OpenShift foundation has the potential to automatically examine x-rays, identify the most valuable diagnostic images among the thousands taken and flag any discrepancies for the radiologist. This decreases the interpretation time for radiologists.
But it also seems to be a big win for openness: Innovation developed internally is immediately transferable to public research clouds such as the Massachusetts Open Cloud, where large-scale data sharing and additional innovation can be fostered. Boston Children's Hospital aims to extend the reach of advanced healthcare solutions globally through this approach, amplifying their impact on patient well-being worldwide.
"Red Hat believes open unlocks the world's potential," the announcement concludes, "including the potential to share knowledge and build upon each other's discoveries. Additionally, Red Hat believes innovation — including AI — should be available everywhere, making any application, anywhere a reality.

"With open source, enabling AI-fueled innovation across hybrid IT environments that can lead to faster clinical breakthroughs and better patient outcomes is a reality."

Thats IBM!

[Joe_Dragon]

Thats IBM!

Analagous to navigation skills

[sound+vision]

I'm thinking that examining X-rays is like examining maps. The more you do it, the better you get. The more you rely on the computer, the less those skills develop.

I'm not against having the tools, but we know what "US Hospitals" are going to do with them. They're going to use it to cut the number of radiologists in half, double the workload, and force them to rely on the automation instead of giving them the opportunity to develop their own expertise.

Lower headcount in the medical profession, and lower skill level. That is how you advance the enshittification of healthcare.

Re:

[gtall]

It can be somewhat of a dilemma if the bot is better than the humans. It would catch more cases but then it might miss cases that humans would catch.

I agree with you that hospitals will misuse the tool and the human skill of interpreting those kinds of XRays will decrease.

Re:

[dvice]

If we ignore the story and what this particular bot can do and just have thought experiment about the problem you suggested.
"If the bot catches more cases than human, but misses some that humans would catch"

Lets say that we examine list of cases that have 100 things we want to find.
- Lets assume the bot finds 90 of those.
- Lets assume a human find 20 of those, but 2 of them are the ones the computer missed.

Question is: Assuming you are one of these 100 patients. Would you like your X-ray to be examined by a

Re:

[Luckyo]

>Question is: Assuming you are one of these 100 patients. Would you like your X-ray to be examined by a doctor or by a bot?

This is of course a false dichotomy. It costs next to nothing to be examined by AI, and it cost an extreme amount of be examined by a doctor. Also doctors being humans make mistakes in pathology ALL the time. Story of "they took biopsy, doctor found nothing, a couple of years later they found metastacized cancer" is not an exception, it's quite common.

And one thing at which AI is not

Re:

[godel_56]

But, this is not actually the only reason why you want to use the bot. Because AI can also, from the same X-ray detect for example diabetes II or predict heart failure or stroke likelihood within 10 years. Something humans did not even think was possible to see from the X-ray. And I think more of these will follow.

Then they'll sell that information to health insurance companies to either increase your cost of coverage or deny coverage all together.

The assistant is replaced, not the doctor?

[drnb]

My understanding from a computer vision class is that doctors make the actual diagnosis. Radiologists assist doctors by examining the x-ray and perhaps high lighting suspicious things. The doctors do an independent look for suspicious things.

The AI based scans might supplement or at worse replace the radiologist's examination. It is NOT supposed to replace the doctor's examination. So the doctor is always the primary scan and the decision maker, and always has a second pair of eyes looking for suspicious

Re: Analagous to navigation skills

[flyingfsck]

The increased and unusual blood flow to a tumour tends to give it away.

Re: Analagous to navigation skills

[flyingfsck]

It is all about pattern recognition and you can train an AI with vastly more good and bad examples than a doctor will see in a lifetime.

Re:

[sound+vision]

So keep all the doctors (radiologists, etc.) and give them the AI tools on top of that. Require some years of experience before letting them use the AI.

Like waiting until a kid understands arithmetic before letting him use the calculator.

I hear we've got all these unemployed people searching for meaning in life. If we're to start addressing that, sticking them in the medical field sounds like a good idea.

Re:

[techno-vampire]

About ten years ago or so, my Primary saw something on my left temple that she thought might be the beginning of a skin cancer, so she had me come in for a Dermatology consultation. I was seen by a resident who had a reference work open to the proper place and carefully compared what he saw to various images, eventually deciding that it was benign. Then, he called in his attending who examined me again, including the comparisons to the reference images and confirmed his diagnosis. Today, of course, the i

Because the incentive is money in the US.

[waspleg]

Not best outcomes or availability for all. The costs are absolutely insane too and propped up by the insurance mafia/healthcare system/drug cartel triad.

This was illustrated pretty well and hilariously in the South Park special about obesity drugs which came out a few days ago (signed up for a free week trial of Paramount+ just to watch it and cancelled immediately).

Re:

[groobly]

It's orders of magnitude harder than map navigational skills. But, yes, expertise comes from practice.

Re:

[Luckyo]

Radiologists and pathologists of the kind that do pattern recognition only (without cutting into the patients, corpses and so on) are going away as soon as we can get proper AI trained. This was known before the current AI breakthroughs.

These fields are really troubled already, mainly because humans doctors are the bottleneck (too many pictures to look at) and the weak link (too many false positives, too many false negatives). Humans are just a very poor fit for this sort of work. We're not designed by evol

This article isn't about tech or science...

[PubJeezy]

This article isn't about tech or science, it's about mass firings of hospital. This article is designed further weaken the healthcare system that Americans depend on to survive.

This article is gross. IBM is doing something awful.

Re:

[NoWayNoShapeNoForm]

As the population gets older in a country there is an observed trend where fewer and fewer people are available to do the work.

As for how many of those fewer and fewer workers go into healthcare, and older people generally need more healthcare resources (human body is not perfect and increasingly prone to "breakdown" as it ages), this is a serious question.

Some countries, like USA, can afford attract foreign talent to the USA through higher wages, higher than those foreign workers would find in their own co

Shortage, due to AI that never happened

[Dan East]

My wife is an x-ray technologist, and there is a major shortage of radiologists due to the hype going back nearly a decade that AI image analysis would replace, or at least significantly reduce the need for, radiologists. That sent a shockwave through the field, and prevented at least some percentage of potential radiologists from pursuing that career over the last several years. Now we are at the point that the lack of graduating new radiologists has really caused issues as older radiologists retire. Our local hospital uses a large radiologist group, and they are now to the point we only get an in-person radiologist a couple days per week (they must be in person to perform breast biopsies, and reading a number of other more specialized types of imagery), and the reading of more routine non-emergency images is backlogged by weeks (even when handled remotely by a large pool of radiolists).

All of this due to the hype of AI based image reading that never materialized (or wasn't ever deemed acceptable by the FDA and other powers that be). Now we're to the point that AI assisted image reading MUST happen to replace the shortage of doctors that the hype over it caused several years ago which is now negatively affecting patient care.

Open, share, build...

[markdavis]

>""Red Hat believes open unlocks the world's potential," the announcement concludes, "including the potential to share knowledge and build upon each other's discoveries."

As long as it is not RHEL itself, right? R.I.P., CentOS. More power to ya, Alma/Rocky.

It's going to be necessary

[RogueWarrior65]

Just look at the quality of people in certain medical schools now. We're going to need a well-trained AI to fix that.