Your verification ID is: guDlT7MCuIOFFHSbB3jPFN5QLaQ Big Computing: February 2015

Tuesday, February 17, 2015

An Analytical Approach to Weight Loss

Lets be honest. I am not big boned or large bodied. I am fat. How fat? I am over 100 lbs overweight and have a BMI over 40. Until recently that would make me morbidly obese, but that classification got change to severely. I guess to make the implications of being obese less stark to those who are obese than to imply impending death.

Recently after attend the Clinton Foundation Health Matters Initiative in La Quinta California. I decided to do something about it. With the support of my family I took some time off and went to the Duke Diet and Fitness Center. I was only there for a week, but the results were good. I lost 15 lbs which is about 15% of what I need to lose. That is not bad for five days.

I highly recommend the duke center. I met a lot of people there who they have helped lose a massive amount of weight. There method works while you are there, and if you adopt there methods with work in the outside world.

That is why am I really writing this. While I was there I become interested in what made patients successful when they returned to their lives and what things could be done to impact those results. I was sad to hear that information was lacking. Instead of people who do this are 50% more likely to be successful than those who don't I heard "we thing this will be helpful".  The fact of the matter is Duke, like all of these places, has little knowledge or ability to explore the data of their patients before and after they are at duke. They are trying, but their only ability to pursue this data is by having patients participate surveys. I already know the method of data collection is fraught with danger and poor in terms of real insight. I want better answers than that and I want them now.

During some of the meeting they recommend the use of tracking apps which they felt increased compliance and improved result. I was happy to see one of those apps was Lose It!. Lose it! was written by JJ Allaire who also wrote cold fusion and Rstudio. He is a brilliant computer scientist and businessman. I downloaded the App.

I was was doing that I noticed the Lose it! license. It allows Lose it! to do analytics on the data users enter into the app. That makes sense, and I believe it is a reasonable exchange for the rights to use the app for free.

It was then that I realized that a lot of the data that Lose it! collects is exactly the type of data that Duke would love to know and analyze. This made me wonder if there was ever collaboration between app makers and medical researchers to use that data to improve the results for patients. The advantage being that the data on Apps is not Hippa protected and therefore more easily explored and typically larger than you could get permission to use in a clinical study.

I need to look into this further.

Saturday, February 14, 2015

R 3.1.2 "Pumpkin Helmet"

Back in October of 2014 the latest version or R was released on CRAN by the R Project for Statistical Computing. I have been an active R user for many years, and I thought that I would pass on a few thoughts that I had.

Unlike in years past new versions of R are released on a annual basis rather then a semi-annual basis in the past. I was concerned about this because I felt this would really allow know bugs to linger of way too long. I was wrong about this. In my opinion, the switch to annual release as contributed not only to the size of the update but to the quality of the update itself. This change was a vast improvement over the old procedure.

In terms of an R users and an probably a very average users. I do not do anything unusual or cutting edge.  So I am not really experience a bug in years. That is amazing statement for an open source platform.

Wednesday, February 11, 2015

Great video on if you do not believe in Sports Analytics your are an Idiot.

This is a funny review of the current state of the sports. The reality is that analytics has taken over the running of sports team. The teams that are good at Sports Analytics win and those that are not lose. The debate is over. Keith Olbermann does a good job of closing the book on the debate using Charles Barkley as an example.



Teams are made up of players whose skills and talent contribute to the success of the team. Most bring those talents to the game they played with heart, drive and talent. What sports analytics has changed is our understanding of what strategies and elements of the game are important in winning games and finding those players that posses those abilities. Not is using a subjective opinion, but using the readily available data. So Sports Analytics has change the ways games are played and what player skills are valued. It has not diminish the value of an athlete or an appreciation of his dedication and skill.

Wednesday, February 4, 2015

The Future of Health Care in America is based on Open Data and Open Models.

The Clinton Foundation puts on the Clinton Health Matters Activation Summit every year. I watched it online last year, and I thought that they talked about some of the core issues facing healthcare in America. This year I decide to go and it was well worth my effort.

Now I am a Data Scientist. The conferences I usually go to are called thing like Strata, Predictive Analytics World and DataCon. At those meeting I am surround by people who do and understand what I do. Not so much at the Health Matters Conference. The attendees there were charities, CEOs of Healthcare Companies like Humana and Tenet along with many administrators working in the healthcare field. It was kind of odd to be so alone, but also refreshing hearing the views of non data scientist on the elements required to improve healthcare both from a cost and outcome basis.

These people are no programmers, statisticians and anything even close. Yet without exception they felt the critical elements to improving healthcare were Open Data and Open Models. That is a powerful statement given that it comes from experts in the field who are not experts in Data Science. However, a truer assessment could not be made. Without the existence of Open Data and the an open reproducible method to utilize that data there will be no system wide improvement in healthcare.

Healthcare today is where US Manufacturing was in the 1970s. Massive, inefficient and producing a low quality inconsistent product. Outside pressure forced them to change they way they did business. They used data to improve their process and started to deliver the superior quality product that their customers deserved on a consistent basis that their customers deserved. As the manufacturing companies were doing this they were also lowering their costs by reducing errors and increasing efficiency. This process had many names SPC, Six Sigma, TQM, Q1, etc., but the methods was the same. People had data and used publicly available approaches to analyze it.

There have been some early efforts to implement this kind of thing in Healthcare before, but they had been pretty unsuccessful or isolated in the effect for a number of reasons. This is supported by research that shows the US Healthcare system wastes $750 Billion a year and Medical Error Death is the third leading cause of death in the US. The reasons for this are many. Primary among them is that healthcare has long been a pay for service model rather than a paid for outcome. Therefore there was no financial incentive for healthcare to improve the health of patients. In fact, it was actually financially better for the providers for their patients to get worse and need more medical care and visits. That is slowly starting to change with the ACA which has within it penalties for hospitals to poor outcomes and incentives for good performance in programs like the Pioneer ACO.

However I am not sure there will be system wide improvement unless the data of patient is made available in its entirety to aid in process improvement and treatment determination. If is it not a process but a guessing game where we are trying to solve the puzzle with many of the letters missing. Also the only way solution get adopted system wide is if the methods we used to find  better way are open to all to understand. This is why Open Data and Open Models are so important in Healthcare. Without Open Data we may not have the data needed to improve the results, and without Open Models we do not have proof that the new way is better.

I really enjoyed the Clinton Health Matters Conference and I look forward to going again next year. Hopefully when some of the things we need to improve Healthcare in America are more available.