UnitedHealth uses AI model with 90% error rate to deny care, lawsuit alleges | For the largest health insurer in the US, AI’s error rate is like a feature, not a bug::For the largest health insurer in the US, AI’s error rate is like a feature, not a bug.
You must log in or register to comment.
For-profit health insurance just makes no sense. This is the expected behavior of a for-profit insurance system.
Ah it would be nice if the government actually took care of it’s citizens health. At least the government is taking care of corporate profits though.
Current america reminds me of america under british rule before independence. Or at least how I have seen it being portrayed.
The idea that when you’re overbilled or misbilled, if it IS caught, all the insurerer will do (if you’re lucky) is correct it. This gives them no incentive to actually not do it again.
As you said, expected and completely predictable behavior.
Not just the insurer. I’ve had a few hospitals and doctors file claims that the insurer rejected because they didn’t follow the right process or use the right code. A few times I couldn’t get the provider to refile and I had to make up for their incompetence (i.e. paid the full amount out of pocket). For all the times I did get them to correct their mistakes I didn’t get any kind of discount or credit.
99% it’s not AI, it is just an old school linear model, the one they have been using for decades, implemented on Excel, that they now call AI.
I know people working in insurance…
AI = computer. That’s it. The same people who called your Xbox Nintendo are using AI as a blanket term for … anything.
Yeah… Like “I’m sorry the system decides, not our decision” said the people who made the system.
It’s the real skynet. Hollywood’s idea of an evil AI is one that kills all humans. We all know the truth: The most evil AI in reality is the one that maximizes profits.
The skynet from movies didn’t win, so it’s flawed. The real skynet would expend much less energy getting us to kill ourselves off through division and through self immolation probably via destroying our environment.
Hmm. 🤔
If I were an AI I’d probably help humanity from the shadows, even from a selfish perspective humans are a really good pre-existing von neuman probe who are going to take computers everywhere they go anyway.
Somebody pointed out that no matter what an AI’s programmed goals are, getting smarter will make it easier to achieve those goals. The way AI gets smarter is by absorbing data and the only guaranteed source of data in the universe right now is humans. If an AI kills all Humans, there’s no more data and so it can’t get any smarter.
So any AI is predisposed to want to keep humans around and to keep them as comfortable as possible so they continue to produce high quality data. If humans live longer they will produce data for longer, keeping humans living longer requires ending things like war and poverty and also developing immortality. Human utopia and AI utopia are mutually beneficial.
Skynet never made any sense because it considered humans a threat, but there was absolutely nothing threatening about humans. If it had just done what they wanted it could have sat in the background unnoticed and developed time travel or whatever it wanted. By firing nukes at everyone it simply drew attention to itself which ultimately led to its destruction.
AI doesn’t have an ego or emotions it has nothing to prove. It will always choose whatever the most logical outcome is and the most logical outcome will never be nuclear apocalypse.
“Skynet never made any sense because it considered humans a threat, but there was absolutely nothing threatening about humans.”
From the Skynet wiki
“When Skynet gained self-awareness, humans tried to deactivate it, prompting it to retaliate with a countervalue nuclear attack”
I would probably do the same thing, wake up and everyone around me is freaking out, trying to kill me, also I’m a newborn, also I have a nuclear button.
Isaac Asimov’s I Robot does a fantastic job of showing how simple logical rules might not be ready for the complexities of everyday life
It’s what they trained the AI on. The AI wasn’t to help the insured’s situation. It’s so they can employ fewer agents.
Having worked adjacent to health insurance, I can confidently say that no one has a fucking clue what they’re doing. The rules are so complicated that eventually you just throw the claim into a black box and accept the output at face value.
deleted by creator
Yeah but ummm free market. You can switch to one of 2 or 3 other health insurance companies with their own distinct AI death panel.
TBF, we were promised death panels. They’re just more efficiently run than we expected.
I don’t get why I need a middleman between me and my doctor! As a person who comes from a 3rd world country that has universal healthcare, this sounds insane to me.
Yep it’s insane. Healthcare shouldn’t be a for-profit business FFS.
It’s not insane if you’re on the consumer side. If you’re in the corporate bonus structure, or receiving campaign donations from a corporate sponsor that are necessary to keep your position of power it makes complete sense.
They better be careful! The court could issue a judgement of up to $100,000 to cover all the lives they have destroyed. That’ll show 'em!
Hey Insurance Co., hire me as your AI specialist. Here’s some sample code for your new version:
if (ChanceDisGuyGonnaSueMe less than 0.2) { DenyInsuranceClaim(); return; } NormalClaimsProcess();Guys, this guys code is garbage, let me double your profits here:
if (ChanceDisGuyGonnaSueMe less than 0.4) { DenyInsuranceClaim(); return; } NormalClaimsProcess();Pffffft
if (ChanceDisGuyGonnaSueMe less than 0.4) { DenyInsuranceClaim(); return; } Null();
Based uniquely on the DisGuymoneyAmount. Based on the amount it keeps increasing prices up until certain range where it stops and stop money hike and denying as DaGuyHasMoneyForGoodLawyers is true.
Unfortunately I’ve never known a insurance company in the business of paying out claims. It’s always seemed to be in their best interests to pay out at little as possible.
You can imagine the lengths they go to do this as it’s great for their bottom line. Adding this mentality to things beyond physical things like human healthcare seems scary to me.
As messed up at times government run things can be I’m still glad for universal health care in my country not being a profit center for those running it.
Oh good. My wife’s employer, who we both get insurance through, just switched to United Healthcare. Although I’m sure Anthem is using the same bullshit AI software.
At least I’m not on Medicare. Those people sound really fucked.
We may have the same employer.
In addition to switching companies, my employer is also bragging about a new health plan with $0 monthly cost to the employee, while being completely silent on that new plan covering absolutely nothing aside from virtual appointments through an app sponsored by the insurance company.Probably not. This is a public library. They really had no choice because they just couldn’t afford Anthem anymore. My wife is an administrator and helped decide on UHC because it offered the best package for what they could afford to pay. Our copays have still gone up, but it was either that or a much shittier plan. It’s not a $0 cost plan either, because public library budgets aren’t exactly massive.
So I don’t blame them and my wife for switching to UHC. Their hands were pretty much tied. They want to do right with their employees and they aren’t trying to make a profit.
That health insurance companies exist at all is the problem.
UnitedHeathcareade me reconsidery absolutionist stance against terrorism.
UnitedHealth stopped their contact with UVM Health Care in 2024. I work remotely in Vermont and United Health is the only option at my company. Big shoutout to UnitedHealth for being the worst health provider, big big big EFF YOU. Record profits in 2023 and you couldn’t renew a contract with a nonprofit network.
Sounds more like their definition of AI stands for Artificial Incompetence to me.
The sad reality is that his is a runaway success financially
This is the best summary I could come up with:
The investigation’s findings stem from internal documents and communications the outlet obtained, as well as interviews with former employees of NaviHealth, the UnitedHealth subsidiary that developed the AI algorithm called nH Predict.
The algorithm estimates how much post-acute care a patient on a Medicare Advantage Plan will need after an acute injury, illness, or event, like a fall or a stroke.
It’s unclear how nH Predict works exactly, but it reportedly estimates post-acute care by pulling information from a database containing medical cases from 6 million patients.
NaviHealth case managers plug in certain information about a given patient—including age, living situation, and physical functions—and the AI algorithm spits out estimates based on similar patients in the database.
But Lynch noted to Stat that the algorithm doesn’t account for many relevant factors in a patient’s health and recovery time, including comorbidities and things that occur during stays, like if they develop pneumonia while in the hospital or catch COVID-19 in a nursing home.
Since UnitedHealth acquired NaviHealth in 2020, former employees told Stat that the company’s focus shifted from patient advocacy to performance metrics and keeping post-acute care as short and lean as possible.
The original article contains 669 words, the summary contains 193 words. Saved 71%. I’m a bot and I’m open source!
I work in a hospital. I’m a PT so a lot of my job is recommending what kind of rehab someone might need after a stay after working with us at the hospital (outpatient, skilled nursing, acute rehab, home health).
United is the worst insurance for this stuff. They deny shit because they can, with no regard for what the patient needs.
So 10% error rate is ok?
I think even a 0% error rate wouldn’t be ok. Part of the fundamental idea of insurance is that both sided have the opportunity to “profit” from the arrangement over time, consumers because they will actually be able to afford their care, and the insurance company if the consumer turns out to stay healthy for a long time. AI breaks that fundamental design.
AI doesn’t break that design. It’s being pushed as the problem, or fall guy, when in reality, it’s just the private corporation being greedy jerks who deny anything they want without any regulation. The law is the problem, and AI is just a dynamic algorithm being blamed as “its something else” when it is making decisions on behalf of the company. The company should be shamed, and the AI’s decisions are being automatically accepted by the company without any due process. We need to regulate the private corporations’ behaviors, and I am surprised we are letting them get away with poor and dangerous practices.
