I might be one of the longer pumpers on the forum (33 years since first pump, an AutoSyringe AS*6 single basal rate programmed by diluting insulin).
While things progressed nicely for a while, since the AutoSyringe/Travenol EUGLY (or Ugly as we called it) there has been no REAL advances! Yes, there are MORE of the same features, more safety alerts, even one or two that will "talk to" a CGMS, but no one is doing anything with the data (except for Medtronics, who just shut down the pump after so many minutes of low glucose with no user action).
Why isn't anyone taking the CGMS and Pump and Meter data, doing some BASIC checking of the data (i.e. did the patient correctly enter his numbers, did he actually calibrate correctly and did he then utilize the values correctly in correction bolus)? Why isn't there software that shows where problems are (multiple days with same pattern of high or low at given time) and then help user to figure out where the problem is (bolus, basal, etc) and aid in finding and adjusting? Why isn't there software to help the patient FIND his basal and bolus rates? This is basic programming: we are NOT talking anything far too complex.
I am currently working on getting some help with this: would love to just write it and test it, but can't get anyone in the pump companies even interested in looking at the idea (they all want that final -do everything- machine that I've been told about, promised, guaranteed and so forth for the last forty years (ever since I was twelve I was already told it was "only a few years away".
Why is no one happy to get something that will HELP ALL OF US on a DAILY and LONG TERM basis??
Any comments, suggestions, etc?
Anyone wanting to discuss idea please freind me and send me a private chat.
I agree that a pump that cooked some feedback/ coaching/ tips would be great. I think the barrier there isn't the pump manufacturers, it's the medical field because *only* doctors are allowed to change pump settings, hee hee. I like the pie charts and do fairly well with them but well, it is a labor. I'm sure my pump could tell me what to do but it's not rigged that way.
The Medtronic Carelink Pro software will make recommendations based on CGM and pump data. My doctor prints the reports and gives them to me. I don’t think we will see a pump do this on the fly until we see a duel chamber pump…JMHO
I think it probably more a issue of priorities and also the FDA. Over the past number of years the CGM companies have spent an enormous amount of resources on improving accuracy of the devices. From my point of view I would rather have them focus on accuracy of the device than number crunching of inaccurate data.
These tools are still medical devices and getting FDA approval is an expensive proposition.
But agree that in a perfect world we would have both accuracy and analysis of data.
I might be unconventional or even odd, but I have been pumping for a long time, too and I do not WANT or expect my pump to HELP ALL OF US on a DAILY and LONG TERM basis. I use my pump to deliver my insulin. And yes, the bells and whistles, and reports, and graphs, and pie charts, and all are fantastic. But in the end, my pump NEEDS me to implement according to my instructions. My pump has no idea what my plan for the day is, it has no clue that I might have an infection, or the flu. It does not know that I might be climbing a mountain tomorrow, or cutting trees. I am not always dependent on technology. Sometimes I travel and leave my technology behind. LOL
There is currently an almost do everything machine that will suspend insulin if a threshold is determined...and I have no desire to slap that thing on me. What if it is WRONG?? It's a machine, and while I admit that the CGM technology is much improved over the years it was a major fail for me. I have all the info I need and will make changes, adjustments, corrections, and have fun for all the rest of my days - I simply don't need anyone or anything telling me what to do and certainly not doing it for me.
Now if it could unload the dishwasher, fold and put the laundry away, chop the lettuce - that would HELP ME DAILY. The long term is in my hands, and also in the cards.
Please do not take it personally but I think you are simplifying the task of pattern recognition. It is one thing to propose a vaguely described algorithm and another to derive solid recommendations from real world data. In our Glucosurfer project we have integrated the latest research like the Amulatory Glucose Profile. We are still very cautious to take this analysis to the level of detailed recommendations.
Take a look at my personal log. Do you see the patterns in the diagram "Glucose - Comparison of days"? Do you see how the diagram "Glucose - Profile per Hour - AGP (Ambulatory Glucose Profile)" is reflecting these patterns by indicating a higher variability for certain periods of the day? Do you really think it would make sense to make textual conclusions to the user in addition to these visuals? Conclusions that would claim a certainty that can never be established? What about all the other factors that influence the glucose: the temperature, the menstrual cycle, stress, work patterns, eating habits and places, type of sports, illnesses and so forth. Do you just want to ignore that by saying do this and that and you are good? If so then this is the perfect explanation why manufacturers of medical devices can not follow this path. They are bound to regulations and can not speculate about the interpretation of data.
The real problem is that pump manufactures follow the typical vendor lock-in. They make it very hard to access the raw data of the pump. Here the FDA should specify that ALL pumps should allow the following:
-can be plugged in via USB -will offer their data as a file system so every OS can read the files
-offer the pump data as a standardized XML file that uses ISO dates and numbers. This ISO standardization will prevent misinterpretation of date values which is a common problem if you process csv data coming from all over the world.
This way our project and hundreds of other web projects could develop tools to upload, analyze and present the pump data. The projects could compete for the best analysis and users could choose the analysis that suits their needs best. Making the data that freely accessible would ignite evolution (the cambrian explosion of pump analysis) I am pretty sure about that.
is it possible for us to speak?? I do NOT believe that a machine should be making the "final" changes to the basal and/or bolus calculations, just give the user and/or care team the best data possible to make choices easier.
We had limited data back in the original pump days 33 years ago, yet we somehow were able to determine a decent basal rate, even with multiple basals and changes of season and what not.
The glucose sensor DOES provide a LOT of data - and sometimes TOO MUCH DATA can be as problemsome as not enough. I am more than happy to discuss and implement possible algorithms that are KNOWN to work, at least for the general cases and we can tweak as we learn.
What is also interesting is how much data we do not even attempt collecting: what is the lead time between a bolus and the bodies absorption/reaction to it? What is the lead time from eating until the blood sugar is raised? How to do different foods, different mixes of foods affect that in the INDIVIDUAL in question??
While we know the basicas for "everyone" I don't beleive that one person actually fits that bill 100% or even 80%!! I think this information may be more helpful in getting good control than much else we know.
PLEASE don't think I've got the cure in mind, as I've gone for far too long hearing about gonna, will, next year, etc. I am just looking to develop tools that will enable us to live better lives easier.
Even a data check on the meter, pump and CGMS for SANITY is missing!! Do you know the % of people who don't utilize the CGMS data when taking a bolus, or who "cheat" and just tell the CGMS that the current blood sugar is.....
Please email and/or contact and see if we can possibly work together on this?
Great idea about getting some standardization. I think pump manufacturers are working hard on a fully integrated system, but I don't think the CGMs are quite there yet.
For me the bigger issue is the cost of insulin pumps and supplies. It's more than a car.
OK, you've heard it before (so have I). But a usable and useful artificial pancreas really will be coming within the next decade this time. CGMs were a critical piece of the puzzle, and those have only been reliable and commercialized for less than ten years. The remaining critical piece of the puzzle is glucagon to pull BG back when insulin has been overdosed - without glucagon tight control would have been impossible without imposing too large a danger to the users. Premixed glucagon is undergoing tests now (they are recruiting test subjects), and once this is available this really is just a programming problem, and a much easier programming problem than the one you are describing. The group at BU have made tremendous strides with their artificial pancreas, read some of the many articles about it START HERE
Sorry I have not gotten back to you on this for a while
I do NOT believe the pump/sensor/meter combo is anywhere near "good enough" to make a decision about someone's care, I am ONLY suggesting that something which would upload all available data, do a "sanity check" on it (look at the papers from latest ADA conference on number of people WITH the CGMS and other good equipment who do NOT properly utilize or even calibrate it). If the ONLY advantage of the system was a) to set all the clocks correctly b) ensure the person is correctly calibrating c) ensure blood glucose readings for their correction bolus d) check pattenrs and overall control for major changes then we have already accomplished quite a bit.
Help user to stop eating in evening for long enough to be dealing only with basal insulin. This is NOT a perfect system (is he ill, is he under pressure, is his system correctly calibrated) but it IS a first step to getting a basal rate that is correct set up. Once that is set, work around the clock in as many small steps as needed to ensure that basal is good.
Determine the lead time between insulin bolus and blood sugar drop (will change depending on the current carb load, the current blood sugar, the general health and happiness of person, etc.)
Determine the lead time between food and blood sugar rise (and this will change depending on types AND amounts)
Using the above, intelligent suggestions can be made to the patient, their care team or their medical team about adjusting how much and when to give meal boluses.
You're claims about the inability to understand the patterns would hold for a person only looking at this patient for ten minutes also: why NOT utilize the information we have to better enable the entire team to work better.
How long have you been working with pumps and sensors? How many years did you do programming on pump UI and features? How much work have you done to improve pump performance and pump/user interaction and correct dosing and living ability?
It would probably be a good idea for us to write off line, as I think my background may be useful on this sort of a project.
I've been a diabetic for over fifty years, a pump user for over thirty years, a CGMS user for over thirteen....
I figured out how to make Teflon catheters (allergy to nickel), worked on getting multiple basals and the concept of a "long term" and "split" bolus into pumps (AutoSryinge 6MP, then the EUGLY) and other more current work.
