_______________________

Disclosure

I am not a medical professional, I have never played one on TV, nor did I stay at a Holiday Inn Express last night.

Posts on this website simply convey the experiences of the author and are not intended to be taken as medical advice.

Talk about any changes you may be considering with your own medical team before changing your treatment regimen.

_______________________

_______________________

_______________________

_______________________

_______________________

Archives

_______________________

Privacy

If you email me, your personal information will not be shared without your permission and your email address will not be sold to any company or entity.

_______________________

Content Rights

We’ve Got DATA!! Uh, So… Now What?

** I’m writing this from my viewpoint as a Type 1 diabetic and, as always, Your Diabetes May Vary

The Atlantic posted an article by Thomas Goetz, The Diabetic’s Paradox, that did a very good and  accurate job of describing why data collection is the arch-enemy of many diabetics, even addressing the psycho-social aspects of the issue.

So let’s just say, for arguments sake, that we now have all this data.  What now?

The analysis. Data by itself, with no analysis, is totally meaningless.

Analyzed data with no action is totally, well mostly, meaningless. Sometimes no action is the correct response.

When you do an analysis, you get results.  Some results are “better” than others, or so we often lead ourselves to believe.  Those results, like an A1c are often “guilt generators”. I, and many others, have often written how easily guilt becomes associated with diabetes.  And I also completely agree with author Will Dubois that there are “No bad numbers, just good data”.

Besides the results, though, the sheer amount of data can be overwhelming! Glucose readings, carb counts, dosage calculations, lions and tigers and bears! Oh My!

I know, that’s pretty facetious, but honestly that is how I feel sometimes when I step back and look at all the simple number of data points involved in my daily care. And decisions based on those data points. How many carbs? What kind of carbs? How much protein and fat.  What’s my carb ratio this time of day? Normal bolus or extended? Making the wrong decision can have tragic consequences.

We all SWAG at times… if we SWAG’d poorly, it is a Supremely Wild Assed Guess.  If we SWAG not-so-poorly, it becomes a Surprisingly Wonderfully Accurate Guess. Some people are better SWAGgers than others.  Personally, I tend to saunter but I have been know to meander from time to time.

There are about a zillion and three things that can influence all those various data points, many of them situational. And to me, those situational circumstances are the ones that tend to get lost in the shuffle.  Honestly, if my meter, pump, and cgms didn’t keep it for me, I wouldn’t have any data to look at.

The large data samples we have really only give me, at least, a couple of ways of examining my data.  If I’m wanting to look at detailed information I’ve found that I can really only look at the last two or three days at most, otherwise the amount of data makes it nearly impossible to keep track of. Also looking at just a few days allows me to remember some of that situational data which I didn’t write down. Things like “I mowed the lawn and got low” or “I guess that chocolate birthday cake was a little more carb loaded than I thought!”.

Where I find my most useful information is when I look at the larger set of data covering the last few weeks, not days. The trick is in how you look at it, you don’t go all the way down and look at each and every meter reading, you look at the averages.  Averages, while being always wrong, can provide valuable information when looked at over time.

Looking at the average of all tests during a certain time-span, say an hour, can be very informational. Stringing 24 of those hour long time-spans together can show you a very accurate picture of how your numbers tend to vary over the course of a day. Understanding that variation is a very important result to me.

In my opinion, an A1c test is a pretty one-dimensional view of my control over the last three months. For example, I once had an A1c of 5.7 along with daily swings from 50 – 350. The 5.7 alone said I was kicking ass, the standard deviation (how wide the swings were) said I was getting my ass kicked.

I don’t sweat the A1c any more, I always want know my standard deviation, to lower it over that same three month time-span. The smaller my swings get, the less hypo- and hyper-glycemic episodes I have, the better I feel, the less I worry and the less guilt I generate for myself. The two results, when combined, give me what, I feel, is an accurate picture of my control for the last few months.

As someone who works in IT and is used to working with large data-sets, it is still possible to get lost in the details.  This is one place where I love how my Endo and CDE look at the “big picture” and help me “see the forest”.

One thing that I have found is that, in general, if it gets too complicated to keep track of, I have done one of two things.  I’m either approaching it from the wrong angle (ie solving the wrong problem) or I have been over-analyzing. My team helps me step back to get a clear look at what is going on, see what the real problem might be.

For example, do I need to increase my basal rate after lunch?  No, I need to change from a normal bolus to an extended bolus because my lunch  tends to be high-protein which means it is digested slower. And keep an eye on the carb ratio…

The devil is in the details, folks. We have all heard that before. What we don’t often hear, or perhaps recognize, is that we really need to understand what level of detail we work best at, at what level we need to work at to get the results best for us.

© 2013 Scott Strange, Strangely Diabetic and http://StrangelyDiabetic.com

  • Stephen Shaul

    Totally agree with everything you’ve said here. I have the same kind of experience with my endo. She tries to get me to see that small changes can make a big difference. Thanks for a great post.

    • http://strangelydiabetic.com Scott Strange

      Thanks Stephen!

  • http://www.mightycasey.com/ MightyCasey

    This is why I’m hoping medicine starts moving in a snowflake direction: IOW, highly personalized treatments built bespoke for each patient. I realize that’s far down the road, but not as far as some might think given the leaps made possible by digital tech and, yes, big data. Fingers crossed.

    BTW, I always thought SWAG was Scientific Wild-Assed Guess. Sounds more, uh, science-y that way?

    • http://strangelydiabetic.com Scott Strange

      Thanks Casey, and yes, SWAG is usually science-y. However, as diabetics, we quickly learn that the science part can only give us a starting point. The rest is based on experience and the current situation.

      Toss in a few other factors that I may not be totally cognitive of like stress level, starting to get sick, starting to get better, or an unforeseen change in plans and it quickly becomes more art than science at times.

      While our treatment is highly personalized, we patients are the ones administering it. I might see my endo a couple of hours a year, which leaves the other 8,758 hours in my hands. We are on one end of the extreme and it takes a lot of time and effort to make it work.

      • http://www.mightycasey.com/ MightyCasey

        Snowflake medicine is indeed an art, informed by science. You’re already a practitioner. I hope the medical industry catches up with you soon.

        • http://strangelydiabetic.com Scott Strange

          I do too, it makes me wonder tho if there will be enough resources available to actually accomplish it. We will definitely need better diagnostic and analysis tools, as many others have pointed out, and they will need to be user-friendly i.e. easy to use and easy to understand

  • http://www.diabetesdaily.com/johnson/ Scott K. Johnson

    Brilliant post, Mr. Strange (it was incredibly hard to *not* call you Dr. there… we’ve talked about that before, right?).

    While I appreciate that we often need to take matters into our own hands in our day-to-day self-care, I really value my doctor’s ability to look at data that I provide him and distill it into meaningful information. That has only happened when I was using Medtronic’s CareLink Pro software, which reflects as much about the priority of Medtronic educating my doctor as anything else. All of the other forms of information I’ve provided have not been nearly as useful.

    And even then, there’s much context that is missing (the lawn mowing and birthday cake, for example).

    Again, great post. Thank you!

    • http://strangelydiabetic.com Scott Strange

      Thanks Scott..

      You hit on a huge issue there, the Carelink software helps capture the data in one spot with minimal extra effort on our part… until other apps can talk to each other, they simply become one MORE thing we have to do. I don’t even look at any apps besides the ones that come with my pump and meter (also Carelink). I don’t use the MM CGMS (simply because that tech simply doesn’t work for my) and I can’t get my Dexcom data into Carelink.

      This lack of interoperability is further exacerbated by the fact that the equipment manufacturers don’t seem to have much interest in having them be interoperable or even in developing standards to allow them to communicate.

      I’m not going to waste my time using another app or device that simply makes me enter data again onto a smartphone so it can be sync’d to some almighty “cloud” where miracles are supposed to happen. That’s not revolutionary, that is app developers not truly understanding the need: the ability to make life easier by having better analysis tools that integrate data for us

  • carolynthomas

    I appreciate your take on The Atlantic piece from the perspective of one who actually lives with T1D every day, Scott. There’s been a wee firestorm over at Linked In’s Digital Health group over this article – but almost all input coming from the tech-hypemeisters blaming the patient for not taking proper advantage of their miraculous technology, or huffy observations about the differences in personal character between T1D vs T2D folks.

    Always nice to hear from a Real Live Patient – and one who even works in IT (bonus!) – about what it might actually be like to live with this chronic diagnosis day after day after day, as you have generously illuminated here for us. Thanks so much for this.

    • http://strangelydiabetic.com Scott Strange

      Hi Carolyn, thanks for the LinkedIn tip, I’m reading through the responses. No technology is going to be used unless it makes life easier. Blaming the patient for not using yet another piece of software illustrates a basic misunderstanding of the market they want to service. We’re already busy enough, don’t expect me to do it twice unless you’ve got one hell of a tool to give me vastly better results

      • carolynthomas

        Hope you’ll add that important feedback to the Linked In comments, too!

  • Kelly D Myers

    If data are tracked and no one analyzes the information, does it make a sound???

    In the last few weeks I have read numerous comments and articles that raise fair questions about the value of self-tracking or the quantified self “movement”. It reminds me a little of the early days of Microsoft and Apple arguing over which was more important, hardware or software. Self-tracking without analysis and learning from the effort is a little
    like buying an early PC and not buying software to run on it.

    The April 1 article by Thomas Goetz in the Atlantic, “The Diabetic’s Paradox”, describes the challenges patient’s with diabetes face self-tracking their disease. Mr Goetz closes with three lessons for self-tracking design by saying it should be effortless, consumer oriented, and address the emotional needs of the patient.

    I agree with Mr. Goetz’s recommendations and I offer two more. Data collection without integration and analysis is futile. I echo your comments, “Data by itself, with no analysis, is totally meaningless.” First, we have to find ways to make it easy for the individual to integrate information from several different sources. Then most importantly, we must give them a tool to aid in their analysis and insights. To date, little has been done to help individuals integrate and gain wisdom from their own EHR, activity, dietary, and medical self-tracked information. We have to give them meaningful and actionable insights for going through the physical and emotional hassle of collecting data.

    New data and ways of capturing them are readily available. The challenge is that all of the data are in silos and cannot be easily integrated for analysis. We have to focus our efforts on ways to identify, integrate and mine disparate data sets to help individuals improve their own health. In biology there is a term called “hybrid vigor” or “heterosis” which refers to an offspring’s superior qualities or vigor that result from crossbreeding genetically diverse parents. Could it be that there might be an “Information Hybrid Vigor” or “Information Heterosis” realized by bringing together individual centric medical history, activity, and nutrition data to find successful patterns of behavior? Can we build an interpretative analysis engine that would recommend small behavior changes that would yield leveraged health benefits?

    The time is now. We need to give the individual the tools to better understand where they are in their disease and deliver daily insights about small incremental changes that can be made to improve their diabetes outcomes.

    • http://strangelydiabetic.com Scott Strange

      Thanks for the comment Kelly!

      You’re right that all the data is silo’d… and much like we’re seeing with EMRs, there just isn’t much urgency on the part of device manufacturers to share with each other.

      You know, the functionality we need is pretty simply and could probably be done with a mostly-simple spreadsheet. The hardest part would be normalizing all the various data formats so it could be added to the sheet in a manner that could be easily manipulated

      • Kelly D Myers

        Scott,
        Agree on the comments regarding the device manufactures and normalization of the variables. It would be interesting to apply some of the more exotic data mining techniques to see if they reveal more subtle patterns…