IBM With Machine Learning at Blue Cross Pilot Program With Machine Assessing Authorizations and Next Level Is Using Predictive A
Posted Dec 07 2012 7:07pm
Here’s yet one more example of pay for performance too and this will in time put doctors in a spot to where they might have to cherry pick patients but that's he same philosophy used by insurers for years. Focus from what I read in this article tend to be more for the payer side than the clinical data and I know there’s an attempt to marry the two here which everyone tries to do but that get skewed with reimbursement algorithms all the time. When the public complain the parameters of what’s allowed with claims may get changed to allow a few more to get paid and when all quiets down the parameters get changed again. Insurers work that way with full on analytics that also monitor the news to give them a time when to perhaps buckle back down again. I heard that from someone who worked doing just that in the insurance business so thus my comments here. I also said early on that some caution was needed with using the technology in healthcare so the non credible doesn’t get tossed in for money purposes and we end up with flawed data.
And the deal is…as someone said on Twitter…we don’t have a Congress that understands how this works sadly.
A year ago Blue Cross made the news with IBM Watson being used at Cedar Sinai for cancer research, who not too long ago closed their inpatient and out patient psychiatry services.
As a matter of fact I said “shame on Congress” for not recognizing a tool that could help them model and make better laws as at least they would all have the same numbers to begin with.
You know that bank named Citi that was reported laying of 11k workers, well guess what they are using…any more proof needed on how big business just runs the gambit with our lawmakers…when they are bliss on technology they can easily be Algo Duped. Algo Duping 101 here. IBM Watson uses Cognitive Computing and “wiseGEEK” here does a good job with explaining it for you and it’s basically replicating the human brain in computer format. When this occurs so much code is getting combined at times that we are writing the unreadable, sometimes without a way to get back to ground zero. Folks smarter than me have also elaborated on this fact.
Earlier this week I commented on the ACO efforts of Anthem Blue Cross and pondered the exact plan of action taking place as it was confusing. Predictive analytics should never be used out of context for decision making processes an do I think that may happen when millions of dollars are out there, could be so we should keep on eye on the math, formulas and codes as nobody watches any of this. To quote someone else (Professor Charlie Siefe of NYU) “consumers don’t like math and thus when shown a formula, they look at it, well gee it has a square root and I’ve read about this in the news several times recently, so it must be good”…happens all the time. I had one doctor write to me and he was too happy with what Blue Cross is doing in the way of compensation for participating in the up and coming California exchange and he’s toying with the idea of doing something else besides being a doctor. So the word is that doctors and hospitals who are already at struggling levels in many areas may disappear so those share holders are happy?
Basically we are taking out time elements as humans and moving it up to that of the computer…if not done correctly and without a focus of ethics, the world clashes and algorithm continue to move data and make more money. The human world and computer world still have to work together and many out there forget that when they have share holders screaming “where’s our money”. Actually I feel sorry for Angela Braly in a way as from what I read she’s was just a scapegoat here. Her individual actions had little impact and the next person at bat will in the same spot. We still do witch hunts way too frequently in business today thinking that CEOs have that much power and control in some areas with big conglomerates. I just love the blatant comment made by a quant on CEOs “they are interesting instruments as they go through and slice and dice”. (Dr. Paul Wilmott-Quant)
To add a little more to this, Blue Cross is under a major battle with United Healthcare as far as competition as Untied has been running analytics and making money much longer than anyone else out there. They are a huge machine and you can pick up some information at this post that maybe you didn’t know about them and insuring people is less than half of their revenue today, rest comes from technology, selling data and so forth. United and Blue Cross are duking it out all across the country and even to having one of the Blue Cross subsidiaries close down after Untied sued to get the DOD Tri-Care agreement. Tri-West announced (Blue Cross subsidiary) announced the loss of the contract is putting them out of business. At some of the links here there are other links that detail all the contract fights and disputes.
Again with the CEO situation the board backed her and shareholder screams and cries for money seemed to be the ticket and again, think they will get a CEO that’s better, chances are pretty much status quo so if you are investor maybe pay attention to some of these happenings and demand changes as it doesn’t guarantee dividends.
Guess what, your dividends are being controlled by the algorithms too, so think about that.
All researchers and philanthropy folks are of course wanting the clinical values to further research so we shall in time if the payer efforts collide there as well. The magical algorithms for a solution are not there; however there’s a lot of data that can make us smarter in some of our decisions and hopefully this will have some kind of emphasis with big data and the use of such with IBM Watson so we don’t get outside ourselves. When reading such articles like this I always want to reflect back on organizations like this one and if we were really doing that good with ethics and handling healthcare like humans, this organization would not exist as we get so involved with playing with risk factors, falsifying them and so forth to make a buck, so let’s do it the honest way with good code for accurate formulas. BD
Our first pilot just went live in five provider offices, where we have providers or their back-office staff sending requests over and our nurses using [IBM's] Watson technology to assess the decision and responding with authorization for procedures. We've made a lot of progress. We're also focused on oncology, and we expect later this year that we'll be into our first pilot in oncology.
We know some basic information based on the data we collect, but as we go more into incentive-based reimbursement for quality care, we're going to have to know more. The predictive analytics we're putting around understanding who are the providers that can best take care of our members in the most effective way -- that's another area we're focusing on because it's tied to aligning incentives and reimbursement.
Incentives are a big topic in health care. A lot of the difficulty is understanding the true quality of providers and the true costs. Can we get to where that's easily comprehensible?
We believe we can. We have a big initiative focused on patient-centered primary care. Part of implementing that work is understanding the metrics, the data points we're going to capture, and how we align incentives with the right quality measures.