SoylentNews is people
SoylentNews
from the Robert's-your-mother's-brother dept.
https://www.hcanews.com/news/how-machine-learning-could-detect-medicare-fraud
Machine learning could become a useful tool in helping to detect Medicare fraud, according to a new study, potentially reclaiming anywhere from $19 billion to $65 billion lost to fraud each year.
Researchers from Florida Atlantic University's College of Engineering and Computer Science recently published the world's first study using Medicare Part B data, machine learning and advanced analytics to automate fraud detection. They tested six different machine learners on balanced and imbalanced data sets, ultimately finding the RF100 random forest algorithm to be most effective at identifying possible instances of fraud. They also found that imbalanced data sets are more preferable than balanced data sets when scanning for fraud.
"There are so many intricacies involved in determining what is fraud and what is not fraud, such as clerical error," Richard A. Bauder, senior author and a Ph.D. student at the school, said. "Our goal is to enable machine learners to cull through all of this data and flag anything suspicious. Then we can alert investigators and auditors, who will only have to focus on 50 cases instead of 500 cases or more."
[...] "If we can predict a physician's specialty accurately based on our statistical analyses, then we could potentially find unusual physician behaviors and flag these as possible fraud for further investigation," Taghi M. Khoshgoftaar, Ph.D., co-author and a professor at the school, said.
So, if a cardiologist were incorrectly labeled a neurologist, that could be a sign of fraud.
Still, the data set itself remained a challenge. The small number of fraudulent providers and the large number of above-board providers made the data set imbalanced, which can fool machine learners. So, using random undersampling, investigators whittled down the set to 12,000 cases, with seven class distributions ranging from severely imbalanced to balanced.
[...] Surprisingly, researchers found that keeping the data set 90 percent normal and 10 percent fraudulent was the "sweet spot" for machine-learning algorithms tasked with identifying Medicare fraud. They thought the ratio would need to include more fraudulent providers for the learners to be effective.
Actually a compelling argument for single payer.
(Score: 3, Interesting) by bradley13 on Tuesday November 06 2018, @10:36AM (3 children)
Serious question - I'm neither for nor against - why would fraud be better under single payer? TFA is analyzing Medicare fraud, so in the context of the study, it already is effectively single-payer data.
AFAIK, the biggest sources of fraud, or just waste, in this context are:
- Incorrectly billed procedures. For example, suppose you have treatment X, which can be charged at different rates. Maybe one rate for a normal procedure, and a higher rate for patients with specific complications. A dishonest doctor might bill the more expensive one even for simple patients, figuring that no one will notice.
- Unnecessary procedures. This can simply be defensive medicine, i.e., do extra tests as a defense against any possible malpractice suit. Or it can be a way to run up the bill.
Of course, there is also blatant fraud, like billing for procedures that were never done. Likely that's a small fraction of the total, though, because it's much easier5 to detect.
More generally in the US: I have numerous acquaintances who have problems with "out of network" providers. One example: A guy's wife is in the hospital, for something not to do with her eyes. Some random opthamalogists pops by, asks her a couple of questions, and later the guy get a bill for the consultation. It was a consultation that (a) he didn't want, and (b) had nothing to do with the actual reason his wife was being treated. Nonetheless, the hospital insists that he pay the bill.
More generally, I have heard of people being admitted to "in network" hospitals, but then all the doctors in the hospital are "out of network", and hence not covered by insurance. From a European POV, that makes absolutely no sense - either the hospital takes your insurance, or they don't.
I'd be interested in hearing Soylentil comments about this - is this kind of borderline-fraud normal?
Everyone is somebody else's weirdo.
(Score: 4, Insightful) by aclarke on Tuesday November 06 2018, @01:07PM
Nobody cares about the kinds of fraud you've mentioned in your last couple paragraphs, as they're fraud committed against the patient. The only kind of fraud Medicare cares about, and by extension the government, industry, and elected officials, is fraud against Medicare.
The point at the end of the summary about a single-payer system is correct in that this should indicate that there's something more fundamentally wrong with the American medical system than fixing Medicare fraud. Certainly with Every.Single.Other.Developed.Country.In.The.World having figured it out (almost not an exaggeration) they don't really have to look very hard or far to come up with better ideas if they want them.
(Score: 2) by All Your Lawn Are Belong To Us on Tuesday November 06 2018, @03:06PM
Yeah, but (always a yeah, but).... Your question on "borderline fraud" triggered something for me. The government did a clever little trick in labeling these things "fraud." Generally speaking fraud means that there is an intentional act to deceive. However, in looking-glass healthcare world the act of billing is always regarded as intentional and so any mistakes made are termed Fraud regardless of their intent. Hence you get things like genuinely defensive medicine, genuinely accidental billing mistakes, and yes, physician mistakes that caused no physical harm (wrong diagnostic pathways that are corrected promptly) labeled with the F-word. You should expect any payer to invest some effort in finding mistakes but ones that are unintentional should always be expected. But ANY recovery by Medicare for ANY reason gets lumped into self-congratulation for stopping "fraud." Anytime I see the words "Medicare Fraud" I reach for my salt shaker to make sure I have a bolus dose ready, because HHS also intentionally inflates that term way beyond reason.
That aside.... to your specific case: What was the friend's wife in the hospital for? Without knowing that it is impossible to know why a provider might consider a consult referral. (Having diabetes for example, often gets you foot and eye referrals. Nothing to do with blood sugar per se, but very frequent complications of having it. Sorry if you knew that.) However, what that means is that there was a physician who ordered that opthamologist to see the paitent and there is a record of that order. (Specialists don't just wander the hospital looking for cases to get involved in). Further, most insurances have LOTS of procedures in place to say, "Yeah, before you order that you have to talk to us first so we can quiz you that you tried other options before requesting that...." Also, at some point that physician had to say to the patient, "So I'm having an eye doctor come to check you out, too." If that wasn't done and documented then the patient never gave informed consent for the consultation to take place. And most Doctors do in fact do that - that's your point to ask "why? what do you expect to find out from that?" A lot of what's seen as waste in the United States are actually physicians being more cautious with their patients than counterparts elsewhere... and some of it is indeed bill padding. How do you tell the difference? And that's why Medicare can't afford the luxury of trying to root out intentional fraud from simple mistakes.
As to network statuses. Yes, what you describe is not completely abnormal. Hospitals in the US do not have firm rules that the providers credentialed to work there are in network with the same insurances the hospital is. (Doing so would violate the principal that physician selection is not based on dollars but merit - a thin veil but one that is real.) There is also no requirement that hospitals credential themselves to all the plans a provider might be in network for. So you can get the opposite - provider has a patient who needs a procedure that for some reason can only be performed at a certain hospital, and that hospital is out-of-network. (Maybe that hospital has equipment or supplies that the in-network hospital doesn't.) Physician services are covered at in-network rates but the hospital is not for the surgery. Every insurance plan has rules in place that requires providers to be actively aware of network providers and make in-network referrals whenever possible. But those requirements are actually requests in practice - a provider had BETTER make most of their referrals in network or the network will tell that physician to shape up or ship out. But there are certainly still corner cases that happen.
This is changing somewhat as "Advanced Care Organizations" and fee-for-value has become more prevalent. Many hospitals now have organizations that are optional for the provider to belong to but pay bonuses or greater reimbursement in exchange for participating in the hospital's cost containment programs. These programs are typically much more rigid about requiring referrals to stay in network (because the hospital has skin in the game and ultimately the Doctors do too.) But such organizations negotiate individually with the payers to contract only special sub-panels and plans of patients and you have to have an employer that picks one of those. Time will tell whether this is actually a good thing or not, because it also hogties the physicians into making medical choices based on insurance networks and not who the referring provider genuinely feels is the best treating physician for a given patient. (Which is also a little bit of a myth - physicians certain do steer patients to certain Doctors. Still there can not be any above-board compensation for that as there are several laws preventing kickbacks.)
This sig for rent.
(Score: 0) by Anonymous Coward on Tuesday November 06 2018, @03:06PM
Worth noting that if you look up false claims act suites, something like 60% of them in recent history are against insurers. In regards to overbilling, there is a simple way to evaluate this. Devide the number of doctors by the number of billing clerks. A typical doctors office has a ratio of four billing clerks to every doctor. I have yet to figure out how it is possible to have 4 people spending 8 hours a day billing for what one man does in 4 hours a day. Unless of course it is wrangling with the perpetual fuckage that is the insurers.
I had a surgery some years back due to a car wreck. I had a 5K deductable zero copay (before obamacare eliminated zero copay policies) and 5K of injury on my vehicle. So I was 100% covered. I recieved a dozen bills, and one bill that was for shit that never happened. Then the health insurer, (Avant) balked on $2k of the bill, claiming overcharging. The doctor billed me the difference. I had to threaten to sue the fucker to get them to sort it out between them, though I never found out what the settlement was. All I got was a revised bill with a zero balance.
These games are rediculous. Insurers don't heal people. What they do is 100% overhead on the services and products provided. Further the state does not compel the insurers to disclose kickbacks. So what you are being billed is in no way reflective of what the doctor is actually charging. So if you're paying a 50% copay, part of that is being kicked back to the insurer, which means that you are paying much much more than 50%. If you consider the actual markup, and the likely margin between the insurer billed rate and the doctors billed rate, in all probability, the insurer actually makes money on most claims, not just on the policies.
The whole system is fraudulent. From end to end. And it actually got worse because of obamacare because the price on zero copay accounts skyrocketed. Which is part of how you know the copay figures are complete bullshit. The differential between the copay and the non-copay accounts are astronomical. They didn't used to be. Obamacare didn't create affordable accounts, it made all the accounts unafordable, because the ones on the lower end, aren't really insurance.