- advertisement -

Artificial pancreas

Discussion in 'Parents of Children with Type 1' started by twodoor2, May 22, 2008.

  1. twodoor2

    twodoor2 Approved members

    Joined:
    Oct 30, 2007
    Messages:
    6,440
    What is the artificial pancreas and the various projects surrounding it? I always hear about it, but my guess is that it's a smart pump that can "talk" to the CGMS to obtain short term BG predictions. Is my description accurate, or is it something else, or much more complex than that?
     
  2. Flutterby

    Flutterby Approved members

    Joined:
    Nov 11, 2006
    Messages:
    14,623
    from what I've hear (and I couldn't be wrong, it happens LOL) is that its the same as the 'closed loop' system that MM is working on.. the pump and cgms will eventually communicate with each other, the pump with have both insuln and glucagon in it.. WE won't have to do anything, it will work like a pancreas SHOULD work.. but I'm guessing thats years away..
     
  3. saxmaniac

    saxmaniac Approved members

    Joined:
    Jun 1, 2007
    Messages:
    4,057
    There's a JDRF talk soon about it here in Boston, the guy developing it has a T1 kid. See here. Last I read, they were experimenting with 2 pumps hooked up together, one with insulin and the other with glucagon. I don't think they were doing CGMS yet, but still using blood pricks. They were feeding those into the two pumps, and concentrating more on the counter regulation.
     
  4. frizzyrazzy

    frizzyrazzy Approved members

    Joined:
    Dec 23, 2006
    Messages:
    14,141
    Here's my question - what would it take to trust this sort of system?? i'd be scared out of my mind.
     
  5. twodoor2

    twodoor2 Approved members

    Joined:
    Oct 30, 2007
    Messages:
    6,440
    I know, I was thinking that myself, I think I'd be more inclined to trust porcine beta cells.
     
  6. dqmomof3

    dqmomof3 Approved members

    Joined:
    Dec 29, 2007
    Messages:
    2,778
    This is exactly why my endo thinks that system will never come to market. Too many variables to account for, especially in children, like growth hormones!
     
  7. vettechmomof2

    vettechmomof2 Approved members

    Joined:
    Jul 13, 2007
    Messages:
    3,037
    I was reading this quickly and thought that you said porcupine cells.:eek:
    Allene
     
  8. vettechmomof2

    vettechmomof2 Approved members

    Joined:
    Jul 13, 2007
    Messages:
    3,037
    Also marsha, what reasons do you have for trusting the pig more? THanks,
    Allene
     
  9. Twinklet

    Twinklet Approved members

    Joined:
    Jun 29, 2006
    Messages:
    2,123
    http://www.jdrf.org/index.cfm?page_id=106675

    Marsha, the Artificial Pancreas project is a huge clinical study that has many, many different facets. Last year, Yale university, in combination with Medtronic actually did a test-run on a closed-loop system. http://www.ynhh.org/press/2008/pancreas.html

    We had the opportunity to hear a lecture from a CA physician whose son participated in this Yale study. He brought slides of his son's BG data while on the device and it was astonishing. I don't think he was out-of-range much, if at all. He told us that the device has a ways to go before it can be available to the general public, but he was very encouraged.

    Last year at camp we heard another speaker discuss the closed-loop system. Apparently there are prototypes of dual-cartridge pumps, and they are being tested with insulin and glucagon.

    The part of the study my daughter is participating in doesn't have anything to do with a closed-loop system. I'll type part of the "Purpose of Research" from the consent form:

    "You are invited to participate in a research study entitled A Randomized Clinical Trial to Assess the Efficacy of Real-Time Continuous Glucose Monitorint in the Management of Type 1 Diabetes. <snip> This study is being done to find out if use of a CGMS that gives glucose readings in real time can improve blood sugar control and help prevent low blood sugar events. In addition, we are evaluating the cost-effectiveness of real time continuous glucose monitoring. <snip> We also want to learn how using a CGMS affects daily life with diabetes. We also want to find out if there are any drawbacks to using a CGMS."

    The very small leg of this study Emily is participating in is to assess whether wearing a CGMS will prevent lows in children with A1C's in a certain range; the other leg assess whether using a CGMS will reduce the A1C of those with higher levels.

    They hope to provide data to insurance companies so that reimbursement of CGM systems will become more widespread.

    The main reason we're doing it is for the free CGMS! Our insurance won't pay for one and I really wanted to see if we thought it would be worth it before we forked over a ton of dough. (Right now we're all thinking it IS worth it so we'll probably pay out-of-pocket when the study is over).

    Anyway, just wanted to share that info with you!
     
  10. twodoor2

    twodoor2 Approved members

    Joined:
    Oct 30, 2007
    Messages:
    6,440
    Thanks Andie, please keep us posted on your daughter's progress. I read the link you sent, and I was particularly interested in this

    The device, about as big as a pager, was developed by Medtronic, a Minnesota-based company. It is an automatic "closed-loop" system made of three interacting synchronized parts: a sensor inserted just under the skin to monitor sugar levels continuously, an insulin infusion pump with a catheter to infuse insulin under the skin, and a computer algorithm that regulates the delivery of insulin based on information from the sensor.

    I would be very interested in how they do this. I'm sure you have to provide settings like the pump to determine insulin sensitivity, carb ratios, etc. . . but how in the heck does it know how to automatically regulate insulin (if that's what it's doing)? Even with a brain, it's difficult to sometimes make these predictions with 100% fullproof safety, especially in young children. I would love to know more about it in detail. Perhaps it's nothing more than prebolusing a specific period of time and looking at BG relative to DIA. I also wonder if they came up with a more accurate continuous function for IOB vs the discrete one that the pumps currently use today. I would also assume some of the settings are more varied and dynamic than they currently are on today's smart pumps.
     
  11. Twinklet

    Twinklet Approved members

    Joined:
    Jun 29, 2006
    Messages:
    2,123
    I don't know the answer to your questions, but it is VERY interesting! The article doesn't mention using Glucagon with this closed-loop system, so it must be a special algorithm that is far advanced than what we currently have.

    Barring a cure, the next few years of diabetes technology will be very interesting!
     
  12. saxmaniac

    saxmaniac Approved members

    Joined:
    Jun 1, 2007
    Messages:
    4,057
    I thought I posted this, but somehow it didn't take.

    Sure, I would trust something like this. After all, people probably didn't trust pumps at first, and we do now! So why not? My FIL has has defribillator implanted in his chest that zap his heart when it gets in trouble. Boy, mess up on that and it's all over in an instant!

    I would bet you a box of chocolate donuts the system is a lot simpler than you imagine - it would know nothing about DIA, carbs, ISF. It would be reactive, not predictive. I imagine the core of it is something like this: if bg is higher than range, give insulin, if in range, stop giving insulin, if below range, give glucagon. Give more or less based on the difference. There are very well-known algorithms that smooth out "jitter" published over the years for variety of technologies. Getting a variable into some known range isn't really that far-fetched. What's hard is measuring that variable!

    Ever study AI or fuzzy logic or neural networks? Then you know that very complex interactions can come from just a very few simple rules. And that's how our bodies are built; not on sophisticated algorithms, but lots of very simple physical rules that interact in an almost unfathomable, godly manner.

    There are machines today that can drive and parallel park cars and machine vision is hard. Really hard. Again, I'm going to guess that it's way harder than this control algorithm.

    The really tough part to this, again me guessing, is a getting good glucose sensor. I think as CGMS get better an AP will become a reality.
     
    Last edited: May 23, 2008
  13. twodoor2

    twodoor2 Approved members

    Joined:
    Oct 30, 2007
    Messages:
    6,440
    I would die to see the algorithms, however, I'm thinking that this is just a closed loop system with an ultra-smart pump, and there will be times you have to feed carbs because it will alarm. As I continually say, no algorithm or super computer can compare with billions of years of evolutionary biology in the form of the cell. I wonder if it calibrates itself based on the BCR?

    I do hope this is a vast improvement over today's pumps and I hope it is accepted by insurance. It would be excellent to have a built in CGMS that is less invasive that "spoke" to the pump.
     
  14. Twinklet

    Twinklet Approved members

    Joined:
    Jun 29, 2006
    Messages:
    2,123
    Scott, you may be right but I'll bet it does have predictive measures in which insulin or glucagon are administered. The current CGM systems all have predictive alarms based on how rapidly the glucose is rising or falling, and we were told to treat that trend. I'll bet the closed-loop system has something similiar but I'm not sure.
     
  15. Twinklet

    Twinklet Approved members

    Joined:
    Jun 29, 2006
    Messages:
    2,123
    If the system somehow includes Glucagon, a tiny Glucagon dose may be administered to head off a downward trend. At least, that is being discussed with the dual-chamber pumps. I'm not sure if that was used in this Yale study, though.
     
  16. twodoor2

    twodoor2 Approved members

    Joined:
    Oct 30, 2007
    Messages:
    6,440
    It could do that if it were to give glucagon, but if not, I don't think so. I don't think that's part of the mechanics behind this since glucagon is expensive and breaks down easily.
     
  17. twodoor2

    twodoor2 Approved members

    Joined:
    Oct 30, 2007
    Messages:
    6,440
    Is there a negative to administering glucagon all the time though? I hear it cuts off the natural supply of glucagon. If that were the case, then the algorithms that Scott mentioned might work if the glucagon amounts worked the same for every individual.
     
  18. Twinklet

    Twinklet Approved members

    Joined:
    Jun 29, 2006
    Messages:
    2,123
    The amounts they discussed at camp were miniscule, but I'm not sure exactly how it would affect liver storage of glycogen. Scientists do know that people with T1D frequently have defective Alpha cells as well, and that does inhibit glucagon secretion.
     
  19. moco89

    moco89 Approved members

    Joined:
    Mar 1, 2008
    Messages:
    2,430
    The artificial pancreas is already being tried out in the UK. Minimed is participating in this trial, even though this article doesn't mention it. http://news.bbc.co.uk/2/hi/health/6110472.stm

    Here's a video about it http://youtube.com/watch?v=i6VYINslF4E

    Also, according to this recent presentation, it takes 3 weeks for the algorithm to adjust to the diabetic's blood sugar fluctuations. Algorithms for insulin delivery-towards the closed loop http://www.easd-lectures.org/copenhagen/index.php?menu=view&id=172
     
  20. twodoor2

    twodoor2 Approved members

    Joined:
    Oct 30, 2007
    Messages:
    6,440
    Wow, thanks for those, in particular the powerpoint presentation by Dr. Dorothee Deiss. I wish I could get a larger powerpoint image of the PID-Controller Automation which delineates the sum of the three formulas used in the algorithm. I cannot see the formulas up close, but it sounds fascinating.
     

Share This Page

- advertisement -

  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.
    Dismiss Notice