This is a very reasonable discussion about big data and at the end of the video the discussion goes into how people do drive imagethemselves off the cliff all the time with data, in other words, same thing I have said here on several posts, if you think there’s an error, question it, don’t drive yourself off the cliff because the GPS unit (as an example) says that’s the way to go.  I have chose a coin called “the algorithm fairies” and yes we see this day in and day out to where folks expect the algorithms to give the divine answer, even without a human interacting.  Think of it, how many times have you seen the headlines that say something to the effect of “an algorithm will be your next doctor”…silly…and yet we see it day in and day out in the headlines.  That’s what happens when you “drive yourself off the cliff” and the media is full of articles like this. 

I just posted yesterday about CMS and the Inspector General with frustrations about the government health insurance site and their frustrations.  What you are seeing here is exactly that in the fact that there’s still some elements of “witch hunting” in here to a degree with making an effort to some short comings but the overall conclusion gathered here if you read between the lines is that the data given to them from the insurance companies is complex and when displayed on the website, “well it’s ready to drive you off a cliff”…think about that.  They are working and included data they did not create and yet are trying to find some accountability for working with “insanity” here to a degree to figure out how to fix it for consumers and in it’s present format, it might not be fixable.  It is what was designed and given to them to work with.  I did make a comment about the one suggestion with the CEO certifying the data for accuracy…in reality what is the CEO going to do, they are not data scientists and have the knowledge, so we try to add accountability to the insanity of solving the solution:) 

CMS and Inspector General Duking It Out Over Government Website Data That Helps Consumers Find Health Plans- Just Wait Until the Exchanges Get Going And We Find Those Shortcomings…

Sure with adding the CEO into the picture we are looking to add accountability to the picture but is this the right answer?  Again coming back to the video, keep the human in the solution, but get the right ones.  CEOs at this point won’t do much more than say “ok” and move on and it’s still a CMS problem to work with insane data as given.

The “dashboard” is not intuitively teaching us about the world of things happening around us and they have to help educate us in a better fashion.   Near the end the conversation about replacing the doctor withimage the computer once again…and he suggests augmentation is a better word than replacement.  Keep the human in the loop and don’t look for the data fairies.  This is part of what gave me the incentive to write about “The Attack of the Killer Algorithms” to where much of the time the human may not be included in the loop and algorithms go nuts at times too.  You have to love the TiVo example in the video too to where it get’s is all wrong for the viewer, with a bit of humor.  TiVo thinks the user is gay and he can’t escape the algorithm:)

Data can fail us too.  It can cause terrible things, i.e. the financial industry, algorithms going off the cliff.  Sub Prime scam is great example of going off the cliff as the entire country went off the cliff as we all were sucked in by the data developed by the banks, which made trillions in profits for them.   imageUsing data in conversations is a good tool to discuss and one of the most important messages is that data cannot be used as a political weapon, read here often enough to where I debunk a lot of that myself with specific issues.  People ignore all of what is known as “reality” and turn and drive off the cliff “because the device said so”…and yes we have marketing in healthcare that tells you to drive off the cliff too, we see it all the time…”yes that formula looks good, it has a square root and gee it’s been in the news all week, well it must be right” even though you are about to drive off a 100 foot cliff.

“People love to put more data on a page”…classic quote here in the video…and it doesn’t make it actionable.  People who learn to interpret it have the tools available.  Technology adds overhead and don’t always intuitively add value.  Should we talk about EMRs now:)  If this was not enough and your nerd skills and reading values are beyond the layman, here’s another great article on models/algorithms ..I just love Cathy O’Neil’s  honesty and the willingness to come out and say “models lie” too.  They do if created to lie and you don’t know the difference.  She’s a data scientist and will crack the nuts for you even deeper than I go on algorithms and when you read this you will totally be aware of the flaws and “guesses” that are used out there today, with selling data that has value…just look at the huge error factors as some of this is just guessing.  Her discussion here is “using a fancy algorithm in order to use a completely stupid algorithm” 

Don’t drive off the cliffs with healthcare big data if you can help it and watch out for the “algorithm fairies” with wings that fly (steroid marketing) that lurks among us too.  BD


The modern discipline of data science understands the world around us through patterns of information. Dj Patil himself coined the term data scientist, which according to him, is the sexiest job of the 21st Century. A data scientist to him is like an explorer, making discoveries in the exotic mountains of data and floods of digital information. He will take us on a journey to the current "hot spots" of the data revolution.  

http://www.youtube.com/watch?feature=player_embedded&v=cuI21oE-Bmk#!

0 comments :

Post a Comment

 
Top