If you are looking for the perfect instrument to start a biological horror show in our age of AI, you have come to the right place. Researchers at Johns Hopkins University have successfully used AI-guided robotics to perform surgical procedures. So maybe a bit less dystopian, but the possibilities are endless.
Pig parts are used as surrogate human gallbladders to demonstrate cholecystectomies. The skilled surgeon is replaced with a Da Vinci research kit, similarly used in human controlled surgeries.
Researchers used an architecture that uses live imaging and human corrections to input into a high-level language model, which feeds into the controlling low-level model. While there is the option to intervene with human input, the model is trained to and has demonstrated the ability to self-correct. This appears to work fairly well with nothing but minor errors, as shown in an age-restricted YouTube video. (Surgical imagery, don’t watch if that bothers you.)
It’s noted that the robot performed slower than a traditional surgeon, trading time for precision. As always, when talking about anything medical, it’s not likely we will be seeing it on our own gallbladders anytime soon, but maybe within the next decade. If you want to read more on the specific advancements, check out the paper here.
Medical hacking isn’t always the most appealing for anyone with a weak stomach. For those of us with iron guts make sure to check out this precision tendon tester!
Definitely. There’s still humans in the loop so the introduction of computer intelligence only adds additional layers of protection against human error.
Same reason why I’d rather a surgical robot than a human wielding a scalpel. Same reason why I am glad autopilot exists rather than depending on a human pilot to be steady and alert without any breaks in focus for 12 hours at a time. Same reason why I would rather use a digital spreadsheet for calculations than manually filling them out on paper.
Clearly you’ve never played a game of broken telephone.
There is a tendency for the autopilot to induce accidents that actively confuse the pilots or cause errors which cannot be resolved by the pilots, leading to the crash, but these are all written down as human errors because admitting that the automated system was the primary cause of the crash would erode public trust and collapse stock prices.
So they always find a way to blame the pilots.
The entire premise that a human surgeon will accidentally cut the wrong thing or whatever is about as close to zero as possible. I’ve worked in ORs for 20 years and seen that zero times. It isn’t that it can’t happen, I’ve only worked at huge centers, but it’s rare. I’m certain someone will pipe up with a “my brother in law’s friend…” story and believe me I’ve heard this too.
More helpful would be to define the margins of, say, a brain mass or something similar. What people also don’t realize is that many cancers (one example) aren’t what you’d imagine with, like, a big weird looking thing that can be cut out. The cancer cells are massed up sure but also in a microscopic scale distributed far and wide. Help
With the risk-benefit of a wide resection via loss of function could be a good AI use.
TLDR- mechanical surgical problems are rare. Decision making is hard(er)
“The entire premise that a human surgeon will accidentally cut the wrong thing or whatever is about as close to zero as possible” ……
Are you kidding me !!????
If it were true, there would be no need for medical malpractice attorneys.
Where do you live ?? Because I’d consider moving there if your surgeons are so perfect.
Either that, or you must live in fantasy land (say hello to Dr. Quincy !).
There are literally thousands, thousands of medical malpractice cases every year.
The cases are blatant negligence on the part of the surgeon.
It sounds as if you have no clue wtf you’re talking about.
Why do you think doctors carry medical malpractice insurance ?
And before someone condemns the attorneys as “ambulance chasers”, if your
“loved one” is a victim of medical incompetence, let’s see how your attitude
changes when a drunken surgeon comes at your family member with a scalpel !
(guy shows up in the OR reeking of alcohol on his breath – nurse stops the
chump in his tracks – cases like this, and worse are documented ad-infinitum).
Let’s not forget that doctor going by the name of Mengele.
OMGodwin’s Law, just had to go there didnt you. Completely irrelevant to the topic at hand.
Doctors carry malpractice insurance because they are required to, not necessarily because it is needed.
In today’s litigious society I am not surprised there are “thousands” of cases a year, but how many are found in favor of the plaintiff (not settled, settled is simply to make an annoyance go away not because a Dr did anything wrong).
To further illustrate the absurdity of your “point” you brought up Mengele. Hardly the case now, and definitely not related to a normal person going to the OR.
Docs do make mistakes, but not very often when operating.
“The entire premise that a human surgeon will accidentally cut the wrong thing or whatever is about as close to zero as possible.”
Not according to actual research: Wrong patient, wrong body part, and wrong procedure surgeries happen about 20 times per week according to a study by Johns Hopkins. (https://www.sciencedaily.com/releases/2012/12/121219111336.htm)
These things happen so often, we even have a term for them: “Never-Events” (a medical error that should never happen, since it is avoidable and preventable.)
That’s correct, the more common error is retained surgical instruments. It’s so rare that it happens “only” 1500 times per year, or about once every 6 hours in the USA.
Out of 310,000,000 major surgeries each year 1,500 issues is as close to zero as it gets.
Next you will claim flying is unsafe, with similar statistics. Did you think no-one would call out the BS?
Based on those statistics, you are 1600% more likely to be sewn up with a sponge in you than die in a plane crash.
And based on the statistics you have a 0.00048% chance of that happening (surgery mishap).
Chance of getting hit by lightning: 1/1,000,000 = 0.0001%
So only a slightly higher chance of a surgery mishap than getting hit by lightning, something that nobody worries about.
Nice try with the “It happens every 6 hours” though
So you’d oppose new technologies to make flying safer on the basis that it is already safe enough?
I object when the system is operating by distributing the risk.
Suppose 1 in 10 doctors are incompetent and risk malpractice. Therefore, if you employ an AI that has an error rate similar to all doctors including the incompetent ones, then you’ll get the same number of malpractice cases and you can argue it’s just as good as the human doctors. Only difference is, you’re not switching doctors every time so you’re not risking the 10% probability of getting treated by the incompetent doctor. With the AI doctor you always have a 10% chance of getting treated by the quack.
Likewise, if you take the accident rate of airplanes, the number of accidents isn’t equally shared by all airline companies. Some few run worse than others, keep worse maintenance and worse pilots etc. If the autopilot you employ is based on merely being better than the industry average human pilot, you’re still increasing the risk of accidents for the majority of passengers.
You sound so sure, but I know someone who was struck by lightning. I’m not sure what argument you’re trying to make here, that rare events are so rare that the people who they happen to can simply be swept away like they don’t matter?
If so, I suppose you would make a good actuary.
The point I am making is that it is not as big and outrageous issue as these two quotes are trying to make it:
“It’s so rare that it happens ….. about once every 6 hours in the USA.”
“Based on those statistics, you are 1600% more likely to be sewn up with a sponge in you than die in a plane crash.”
They are trying very hard to make it seem like a huge issue that the actual numbers don’t support.
Yeah close to zero…
I lost two uncles due to mistakes like this. One uncle had his gall bladder accidently sliced and the doctor “never saw it”. He got home, few days later he was back at the hospital, spend 6 months in the intensive care unit until he died. Another uncle died because of a prostate biopsy and the doctor accidently poked the needle in the wrong place, which caused bacteria from his colon to go into his nerve system which ended up in his spine. He spend a year in so much pain not even morphine could help ease it. He died as a result of it.
I’ve seen so many doctors make mistakes. Friend of mine had an accident with one of my motorcycles. The doctor that looked at an xray of his hand missed the 28 broken bones inside his hand. Week later he had to get surgery that took a lot of hours, tons of metal parts got added to fix parts of his bones that were just laying there not attached to anything. With me, they missed 4 fractured ribs with x-rays. Went back to the doctor after a month of pain, another doctor looked at it and noticed the fractured ribs.
Doctors make mistakes and it happens A LOT.
“estimated that >200,000 patient deaths annually were due to preventable medical errors.” And that’s only in the United States. One country, 200.000 patients that die due to medical mistakes. It’s a serious problem.
Preventable includes things like “cant get an appointment quick enough” in fact that is most of Canada’s yearly preventable deaths.
That might be true for Canada, but considering the plans of Canada, I wouldn’t use them as an example, unless it’s a horrifying example. And if bill C62 (aka Aktion T4) becomes active…
I am a military vet so I am acutely aware of medical via a Canada like system…. Sadly its what people in the US seem to want without realizing how terrible it is.
I would rather be broke than dead.
Reminds me of a term I heard while traveling up in scandinavia. The Finnish pun for a public state funded GP is “a guessing clinic”.
I agree the odds are low and having a surgery has never really bothered me beyond the fact I was about to get cut open. That being said, developing new technology that has the potential to decrease the odds of malpractice even closer to zero is an endeavor worth exploring.
The only thing more confident than a surgeon is a LLM surgeon.
Ah this will come in handy for DIY stroggification.
Some day perhaps …….
Dr. Beverly Crusher: I swore I’d never use one of these. Computer, activate the E.M.H. program!
Emergency Medical Hologram: Please state the nature of the medical emergency.
Dr. Beverly Crusher: Twenty Borg are about to break through that door! We need time to get out of here. Create a diversion!
Emergency Medical Hologram: This isn’t part of my program. I’m a doctor. . .not a doorstop.
Dr. Beverly Crusher: Well, do a dance, tell a story, I don’t care! Just give us a few seconds!
(The Borg break in.)
Emergency Medical Hologram: (attempting to stall them) Uh. . .according to Starfleet medical research, Borg implants can cause severe skin irritations. Perhaps you’d like. . .an analgesic cream?
I don’t trust anyone with surgery, human or otherwise. By nature, it’s a desperate roll of the dice. Your body is best left intact unless you are actively dying.
A desperate roll of the dice? Youve got to be kidding.
There are hundreds of conditions that can be corrected with minor surgeries that ignored lead to worse conditions, lowered quality of life, and needless suffering.
Do you hold the same opinions of medication? The risks associated with many medications are often as great or greater than those involved in surgical interventions.
I do hold the opinion that surgery never makes things good again. It might fix the problem, but it also leaves its own problems behind. Scar tissue, slight nerve damage, blocked off blood vessels, things like that.
Like when I broke my leg. Surgery was necessary to repair it, but also damaged a nerve and caused scar tissue inside and outside, which affected my walking and is now causing me all sorts of problems with my hip and lower back 30 years down the line. It had to be done, but it caused its own damage, and is limiting me now.
So surgery should only ever be done if it’s absolutely necessary.
For instance “I think that if I operate this little thing, your problems might go away. Let’s give it a try, why not?” is a very weak excuse for surgery.
Surgery is not the main goal. Surgery is done to achieve that goal. And for that, you first need to fully understand the problem, fully understand the solution, and make a fully fool proof plan to reach that solution and fulfill the goal.
The goal is to become better, the solution might be surgery, but might just as well not be.
“For instance “I think that if I operate this little thing, your problems might go away. Let’s give it a try, why not?” is a very weak excuse for surgery.”
I dont knoww what country you are from but Ive never heard a doctor/surgeon present treatment options like a 13 year old boy shooting for his first date like that.
If your leg break required surgery you were going to be a lot worse off than you are without out it. I cant even begin with you. Enjoy your Bliss!
PS stop voting please
It is quite common. Surgery is preferred instead of no action because it’s doing something rather than nothing. Over time, many common surgical procedures have turned out to be unnecessary in the sense that they don’t provide any living quality improvements for the patient, or they’ve actually made things worse over time.
For example, attempting to repair the torn meniscus of the knee in elderly patients is almost always useless and physiotherapy would reduce the pain symptoms better in the long term according to modern research, yet it is recommended by doctors and performed regularly. Old habits die hard.
Or this: shoulder injuries by home treatment instead of surgical intervention.
https://pubmed.ncbi.nlm.nih.gov/29433644/
“Approximately 75% of patients remained successfully treated with nonoperative treatment at 5 years”
There is a definite bias of “let’s operate” instead of “let’s wait and see what happens” – not for the smallest part because the patients expect and demand something for the money they’re paying.
RFK, is that you?
No, I just know a surprising number of people who have been killed or injured by incompetent surgeons while being treated for non-life-threatening conditions, and know better than to try my luck.
I just experienced robotic surgery. I have to credit the surgeon, but my results were much better than anything I’d read about on the internet and the entry wounds were small and healed fast. Two months after the organ removal surgery, I’ve little evidence that anything occurred while others that had similar traditional surgeries recently are still having major issues and complications. It is indeed a new world we live in…
I wouldn’t want to discover that I’m the first person that the AI has seen with inverted internal organs.
Hey this guy has inverted organs, GET ‘IM
Nah that’s easy. You just need to flip the encoder cables on all the robot arm joints upside down.
for me you can leave out the last three words and the answer will be no. A llm has a opinion at best.
Nope.
Nope, nope, nope, nope, and nope.
There’s no way I will ever trust an AI to work on me when under the knife. Heck, I don’t even trust them to transcribe an email for me correctly!
With all the problems AI have today it already is better than human in many cases. That is why you have a trained personnel to guide it. And I think this is going to be a problem. People soon will assume that you don’t need that level of training because you have AI assistant today and this will lead to degradation of operator.
Basically the whole problem is that people mix LLMs with AI. Helped by the media and the guys who make LLMs, because the term ‘AI’ sells better than ‘LLM’.
But an LLM is not Artificial Intelligence, it just looks like it is.
“Any sufficiently advanced technology is indistinguishable from magic.” – Arthur C. Clarke
For us engineers and hackers, the technology is not that advanced. For the other 7.8 billion people on this earth it is.
“AI” or LLM is autocomplete + Google search in conversational form.
LLM is a subset of AI. The “M” stands for “Model”, as in an AI model. Kindly read a book before spouting garbage.
I’m an engineer who actually works with AI and has for a few decades now. If you don’t fine it advanced then I would love your help optimizing a model since the greatest engineering in the field today find it pretty darn advanced. Maybe drop a link to your GitHub and let us see what you’ve accomplished?
I mean my github is mostly radio stuff, with major feature contributions and fixes to the major radio projects. Some of my personal projects I haven’t published because it could be easily abused (recreation of the Saflok python detailed at DEFCON but unpublished), keeloq stuff, a majority of my Yardstick One tooling (rolling code interception and decoding + autoincrement tx) and a variety of other mostly security focused projects.
Please explain how AI “thinks” because as far as I am aware it simply tokenizes inputs, evaluates weights defined by the training dataset and returns a probability based response. Which is also why it is f*cking terrible at writing code.
And yes I am a Software Engineer by trade.
I don’t believe that “intelligence” is in any ways related to LLMs. It’s just modeling the properties of language by calculating its statistical properties and replicating that in a stochastic manner.
I mean, it fundamentally amounts to throwing dice to select a book and a page from your library, and quoting whatever is found. Where is the intelligence in that?
Dermatology is one of those fields. Software can detect in photos early problems that humans can’t.
Rather than 15 minutes with a physician’s assistant who doesn’t know me looking at an incomplete medical history because when the US went to electronic medical records the people in the doctor’s office didn’t enter most of mine (too lazy and graphs/readouts may not have been possible to enter) and handed me a stack of paper of what they didn’t, I’d rather have a specialized medical Watson who has read all of the latest medical studies (rating those from big pharma with near zero credibility/weighting for proven reasons) diagnose me from a COMPLETE medical history. It would also give options based upon costs related to an individual’s specific medical insurance coverage. The diagnosis and recommended treatment(s) would then be reviewed by a human doctor.
Forgot to add that it would “show its work,” giving the reasons it came up with the specific diagnosis.
It’s probably better then the Swedish Cheff I guess.
Todays arrogant surgeons, with early Parkinson symptoms, that refuse to retire and the gready surgeons, high on amphetamines, operating people like an assembly line for profit scary me more than a future AI robot surgeon supervised by humans.
I once had a dentist that drilled in healthy teeth and filled them to make more money. I think he eventually served some jail time or fines but I’m not sure.
My mother went in for a root canal and the dentist saw she was diabetic. He concluded the record must be wrong because skinny people can’t be diabetic. He told the anesthesiologist not to use the protocol for diabetic patients, and after the surgery she wouldn’t wake up. She was in a coma.
At the hospital where my father worked IT, a kid was going for surgery. Before the operation even began the anesthesiologist pushed the wrong medication into the IV, injecting him with a local anesthetic instead of general because they hadn’t bothered to read the labels. He died as soon as the drug hit his heart.
Myself, it took 3 doctors before one took me seriously enough to believe when I said I was having heart trouble. The previous doctors said I was too young to have any heart problems and it was just my imagination. Finally found out I have an arrhythmia triggered by stress and stimulants and had to make some lifestyle changes to accommodate it.
I absolutely want an AI double-checking everything a surgeon does inside of me. And every dentist, anesthesiologist, and just plain every doctor. An AI isn’t trying to finish in time for 3pm golf, a doctor is.