We are hearing more and more about the use of machine learning and this week we have WellPoint banking their bucks on reducing costs with it’s use. The computing power is phenomenal and if you are not already aware, there are other industries that are using machine learning too. You will get more benefits by a landslide; however a lot of this is brand new territory and we have seen some oddities that have occurred, one being called a Flash Crash and we all know that one. You 
have some algorithms that are programmed to hide and others that are programmed to go find them.
The Quiet Rise of Machine Learning-Like IBM Watson Offers With Speed and Learned Analytics-Congress Needs Technology This To Make Better Laws & Collaborate
The reason for caution is that there are going to be some odd happenings with data here and there and thus the awareness to question if something does not appear to be correct. We have so many digital illiterates today that don’t do that and never question and other people who are at the end of their decision process get hurt or fleeced. Again, the data being input has a huge impact as well as the old saying of garbage in and garbage out has not changed and I don’t that part will ever change but we are getting better. Microsoft has a tool kit for writing “machine learning” applications too so everyone’s into it and it works of course with the cloud.
Microsoft Research Launches Free Tool Kit to Assist Big-Data Scientists-More Ways to Use the Azure Cloud With Large Scale Data Sets And Algorithms For Analytics
Wait until le load up those ICD10 codes in there to decipher 
whether or not that bite on the patient was from an Orca or a McCaw. There’s a ton of processing power with IBMWatson 70 servers…as I said lawmakers could really make good use of this and they need it more than Wellpoint but you do have to pay for it and lawmakers could behave like real humans again and collaborate if everyone starts with the same numbers.
WellPoint Hiring IBM Watson Technologies–Congress Needs to Wrap Their Heads Around Using Technology That Processes 200 Million Pages of Info in 3 Seconds to Make Laws
This is a great presentation by Kevin Slavin who tells us about the affects of algorithms in the world today and worth a listen. 3 years ago when I put the word center stage on my blog, it was for a reason to educate and bring the word and it’s meaning into every day use in healthcare. I had people telling me I was nuts and nobody’s going to get that,…hmmm, but where are
we today and yes we are writing the “unreadable” as we have known it for years with software development. You have to love his example of 2 algorithms on Amazon that with no human intervention but just reacting with each other, ran the price of a book for sale up to $12 million, just out bidding each other and stuck in motion. This is why I say caution with healthcare and to question something that may not look right as we will need to live with those infrequent algo mashes that go bonkers, like a Flash Crash with those algos. The video is below but use the link for even more information I included on the original post, lots of it there.
How Algorithms Shape Our World–TED Talks Kevin Slavin–Writing the Unreadable And A Good Reason to Get Into the Math
Now how monumental and important is this, well if you have seen the latest Ipad commercial, there’s Kevin Slavin’s talk right at the beginning focused. Check it out below. Truth of the matter is we don’t have the luxury any more to say “I don’t do math”, not my doing just bringing around the awareness.
And speaking of books, I’ll note this one as a must read again as I have so many times on this blog…maybe I’ll get royalties <grin>, just kidding there. I have folks on Twitter telling me they are reading now, folks in healthcare. Deceptive, mismatched data and algorithms created for marketing versus actual accuracy lurk out there today sadly and we are actually seeing some of them become unraveled as people ask questions, “show us your code. “
This is absolutely huge in the financial side of things right now with FINRA and the SEC asking high frequency traders for their code. In the High Frequency World of trading the battle of hardware is just about coming to a close, in other words there’s not a lot more they can do to make servers go faster, so the next move, are more “creative machine learning algorithms” for the coders to write and believe me, they are on this already.
“Proofiness–The Dark Side of Mathematical Deception”–Created by Those Algorithms–New Book Coming Out Soon
Back on target, yes IBM Watson will give us some huge abilities to sort and analyze data quickly, but we will need to be aware of those rogue algos as they will pop up here and there and there and we’ll study them just the quants do on Wall Street that pull those algos out of the exchanges to do the same thing. It is critical too in healthcare as we are now talking a human life and thus the words of caution to ask if something looks strange and don’t let your common sense get buried with results on the screen if they don’t look right. Here’s some other folks with some predictive machine learning to market.
An Algorithm Can Predict Cardiac Arrest 24 Hours Before it Happens With “Machine Learning Technologies”
With involving the intricate and confusing coding of healthcare and if and when this develops beyond just guidelines and individuals and doctors are “required” to use this information, we could very well be in the “land of no return” when it would come back to tracing a claim and how the diagnosis and CPT payment codes are entered. Some of course, the obvious would be there, but what if? 
What if those algorithms with claims numbers, diagnosis numbers, and other related information went haywire…would one end up with a medical bill that got inflated just like the example in the TED video of $12 million for the book on sale at Amazon?
Could happen but we don’t know yet so again and awareness and the fact that there will be issued to live with when machine learning becomes larger in healthcare just as it is in the financial market and we can get some Quants in there to work on them, just just like Wall Street does.
We can start naming our own healthcare algorithms, like the “ICD shuffler, the surgeon’s knife or maybe CPT Chaos”, and so on <grin>.
The algorithms have teeth with contemporary math. So again, let the doctors test and try and test again and from the consumer side, guidelines only and don’t tie in in pay for performance by all means with the use of machine learning or it will be abused and mis interpreted for profit. BD
IBM demonstrated how doctors can use its Watson supercomputer to diagnose and treat patients at the National Press Club on Tuesday.
Dr. Herbert Chase, a professor of clinical medicine at Columbia University, entered a hypothetical patient's symptoms, and Watson was able to correctly diagnose the patient with Lyme disease.
Watson even suggested a drug to treat the condition. Dr. Chase then told the computer the patient was pregnant and allergic to penicillin. This new information led Watson to change its suggestion to a more appropriate drug.
0 comments :
Post a Comment