Here we go again with a little different format with trying to answer and create questions to answer in this latest effort with a new carrot. This project is a little different from the original one created to create the “ultimate” algorithm to predict and help drop re-admissions to hospitals that I think is still going. We keep side stepping work with ether bad or erroneous models and try to make sense out of them, so the circle continues and here’s a spot to where you can spend even more time if you want to submit questions while the group tries to figure out what to do, I say get better models than waste a lot of time here. Sure there’s a potential for a lot of questions and it will keep you busy, but do you have time for this? Again I was not very keen on the original project as again it was more of a pie in the sky project and once in a while I would look over there and it resembled a gaming site in my opinion. Sure there was code being written, but for what? It’s this non linear data chase thing that has captivated the country. Quants and modelers know better for the most part. In particular with the banking industry those hired to write models too are feeling the pressure of some of this as they created their own code of ethics so as not to be coerced in to writing models that have high levels of errors and don’t work or possibly lie.
So What’s Going on Over at the Heritage Health Prize Competition to Predict Who Ends Up Being Re-Admitted to the Hospital–A Lot of Mathematical Algorithms For One…
Heritage Providers Continues to Promote $3 Million Dollar Prize to Create An Algorithm To Predict and Prevent Hospitalizations
I guess this site will somewhere along the line make a little money but how I have not figured out since it is brand new. Again being I blog about the business of healthcare, I see this stuff come and go. We already have “code for cash” (link below) going with a lot of big companies, (write apps for our platform) out there. Sure there are participants in some of this hoping to do a start up company and not very many make it, but maybe they wrote some code that the sponsors like and can use, so the developer gets a few bucks and the big corporation benefits without having to hire more developers..the code for cash phenomena. All of this I think got started when the X-Prize and Wellpoint sponsored their 10 million dollar prize a few years ago, the carrot being to award a prize for solving healthcare. Where are we now? Their goals wasn’t possible as we wouldn’t have this new carrot arriving today if it were even feasible and was more of a publicity stunt really. Here’s a bit on how the “cash for code” works. If you read the news today you hear it all the time on how companies are sitting on $1.7 trillion in cash, the latest number quoted by Bill Moyer in his report.
Verizon Latest to Enter “Code for Cash” Prize Format With $1 Million Top Pay Out for Writing Healthcare Apps That Use Their Platform, Is This the New “Corporate Business Model” To Yield Inexpensive Code?
I looked a the prize board here and some of the same and some new players here with former HHS director Leavitt in here again and former US Chief Technology Officer Aneesh Chopra. He was good during the Presidential transition as a real cheerleader to get everyone interested in technology for sure but as time wore in it was “get rich writing code” that seemed to permeate from later speeches and sure people need to get paid for what they do and be motivated but toward the end it lost it’s focus and I would have rather heard more about doing something that helps mankind in the forefront. I see a bunch of other people on the prize board too that seem to be lacking where the use of models is concerned and sure it’s complicated with math models but again to me this is another side effort here and we can wait and see if it takes the same course as what we have already had out there as I discussed above.
Three years ago I made this post with a bit of satire about data addiction and gee, sure likes that 12 step program might be needed soon, either that or have more lean about modeling and see the reality of data mechanics:) A good book on that topic that helps explain some of this is “Proofiness, the Dark Arts of Mathematical Deception”…even well known author Tom Peters and I chatted about it and he went out and bought it. We got on that topic after agreeing on Christopher Steiner's book about Algorithms taking over with “Automate This”…good video here from Steiner and you can find it on my Algo Duping page, which is a collection of videos done by people smarter than me that educate.
Data Addiction and Abuse –The Up and Coming Next 12 Step Program Is On the Horizon–Side Effects Include Lack Of Data Quality, Integrity And Spasmodic Algorithms
In this new “carrot” program we have a Q and A format without a lot of algorithm searching this time it appears. Who’s going to ask a winning question and what data scientist is going to model it? Is this the course of action with all of this? Who cares is my opinion and I’m sure I made no friends with that comment, but really who does care? I guess once they have a “winning” question, chasing algorithms and models is the next step? I ask that question as I’m trying to figure it out myself, but I come from logic and understanding data mechanics and how they really work, models and algorithms so to me this looks illogical:) The members of the board don’t appear to be data scientists themselves but want to give questions to data scientists, who today are working to try to find value even more data? I wouldn’t even mind some data scientists commenting here on this as well.
As a casual reader, if you don’t know what a “model” is, I suggest this video as it relates more to the financial world but the same “model” concept applies with healthcare especially with those wanting to profit. Quants, the Alchemists of Wall Street is the only video I have found that explains this in a format to the layman as to what a “model” is. You have Paul Wilmott at the blackboard showing you, none of what else I have found out there does this. He also dives in with the fantasies of non linear data which is what all the “modeling” talk is about and the error factors when we move from a non linear environment to non linear. Just key in on those two words and the concept and you don’t have to be a mathematician to understand this. In addition, you have to key in on the comments from Mike Osinski who wrote the sub prime software that all the banks used for all the loans, ( NY Times 2009 story at the link and good background information here) he just wrote it but had not control on how the banks used it and we all know what happened there. The best quote of the entire video":
“You can do anything with software” and he continues to say how a lot of this doesn’t work in the “human” world. Myself I related a bit to Mike talking about the hours and hours that goes in to writing software that others don’t see, I have because I have done it. It’s true and “you have to be right” he also says, and that is true too and extends way beyond just financial software, it’s everywhere. So anyway I’m looking at this “prize board” and guess what, doesn’t appear to be any quants or modelers on it for this prize but a lot of hype about data scientists for sure. I made this post a couple weeks ago and the video I used fits here and well worth including about folks driving themselves off the cliff with data because if you don’t have some hands on with data mechanics, people do get duped and drive themselves off the cliffs, see it all the time.
Big Data Revolution-Dangers of Using It To Drive Ourselves Off a Cliff -Debunking Myths of Algorithms Replacing Humans In Decision Making Processes in the Healthcare Business (Video)
Even CMS is having issues with this when working with the complexities of models built by insurance companies. You can hear the frustration loud and clear with the Inspector General wanting the CEOs of insurance companies to “certify” their information they provide for the Medicare insurance shopping site. They are frustrate with the complex models built by insurers that yield profit as a priority.
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…
Again in writing this blog I try to educate and help where I can show the difference between myth and what can or does work. I said a while back that data analytics used out of context stands to be a huge attack on consumers as it will and does get used out of context and the “data selling epidemic” even adds more to the clouded proclamations of value. The link here has a good video with companies talking about how they use data and how to work with Quants. People create “models” that are not verified or questioned to sell things. I have had a couple discussions with folks at the NISS on this too and they agree, there’s a lot of junk out there. Nerds and modelers are never in control and listen to this video from a former Hedge Fund Wall Street quant and she tells you straight out how it works and spent some time doing the analytics work for Larry Summers…good video here for sure which tackles modeling for inequality. I agree with this group of Australian bankers as well, that half of the analytics produced will a waste of money as the data scientists themselves are still working on trying to find value so choose your analytics carefully. I asked one of the folks at Forbes last week on his column if his article was “real or Memorex” as the site uses an algorithm to produce some of their news:) I like this quote from the video mentioned above and it’s true.
If you want to look at data being used out of context, look at this company that doesn’t even bill themselves as a credit scoring company, but that’s what they do at e-Scoring and you can’t get a copy of what they sell about you…
In the video above he talks about “data vomit” where “people like to put more data on a page”, pretty funny comment but real. Technology must teach you to learn he says. The fallacy of computers replacing doctors is discussed here with the doctor interpreting the data, keeping humans in the link. How many times and you see almost ever week about computers replacing doctors, oh I get tired of those articles.
In summary after reading this entire website they constructed all I can say is what in the heck are they doing here and do they know themselves? It sounds like a bunch of noise to me. I don’t see a purpose at all but you might see differently but I use logic, not emotions and politics with my opinions and always keep the human in the loop as I do include ethics with my logic. As the video says, “data can fail us”…the error factors that comes back to having a good model so again for the life of me I can’t figure this site at all with what their real goal is. Some data gets used as a political weapon, we see it all the time, driving off the cliff when common sense tells you something is not correct but people drive off the cliff anyway assuming the data is always correct. Again this brings me back to models and Algo Duping. Great videos to help understand some of this as models are built that are not honest and telling the difference is hard as there’s so much steroid marketing out there (which is kind of what I see here with this site). Again, open for any comments as ethics, data and looking for Algo Fairies lives amongst us today and this is what inspired me to begin the Attack of the Killer Algorithm series to show where data is not interpreted properly for one reason or another and how it costs the consumer largely with having to fix all the flawed data that has arisen on their own ticket and corporate USA thrives on billions in profits, selling it. Scroll on down and watch the 4 videos at the bottom of this page if you want to understand more and most of the material in the videos is at the layman level. BD
0 comments :
Post a Comment