Diabetes Patients Are Hacking Their Way Toward a Bionic Pancreas

There was an insightful reddit comment about this:

>As someone who sells insulin pumps for a living, in my opinion this article is slightly misleading. It makes it sound like the technology isn't there to do exactly what the father/boyfriend want to accomplish: essentially they want to obtain tighter control for the son's/girlfriend's diabetes. To be clear, we do currently have the ability to provide patients with a continuous glucose monitor (which is relatively accurate), and along with that, a model of pump which will stop insulin delivery if a patient's interstitial glucose goes too low. What we don't have (that's described in the story) is a pump which administers insulin doses automatically. But the reason for this is not because of lack of technology... It's because of liability, and that's an important distinction to be made. The boyfriend in the article who developed the auto-delivery system using algorithms for bolus insulin amounts is not doing anything new. In fact, most insulin pumps already have the ability to calculate insulin doses for patients. Instead, he is just doing something that the FDA will not approve of at this time. Every single diabetic responds differently to insulin, and formulas can not necessarily take those variables into account before insulin delivery. As a result, the amount of risk involved at the patient level is too great for the FDA to say "ok" at this time. Additionally, the insurance companies have no motivation to press on because of how costly the approved technology could be. In any case, the article is informative and does a really good job explaining in simple terms how diabetes works. I do believe that we will see a manufactured cure within the next few years (I.e. artificial pancreas) and potentially an organic cure within the next 10-20. Those people who have diabetes deserve that technology and I am confident we will provide it soon :)

As a 19 year old hacker, who has been diagnosed with type 1 diabetes for 8 years, I always had the dream of continuous blood sugar monitoring. But in the real world, a lot of sick people don't have access to these opportunities. The sensors that monitor blood sugar are expensive. I don't mind checking my blood sugar 12 times a day, I hardly feel a thing anymore. The problem is that the insurance only pays for 4 checks a day. If they really want to solve the problem, they should implant the sensors with a surgical operation. Moreover the catheters(infusion sets) we have to wear for our insulin pumps are often faulty. When the needle hits a muscle tissue, it doesn't work, and you are subjected to high blood sugar levels and frustration. Finally, I have to add that, diabetes has been there for years. I was a child when I was diagnosed and my parents never came to check on me every two hours. After a while your survival instincts kick in and you start to feel when something is wrong. As a person who experienced a range from 22 mg/dl to 798 mg/dl consciously, I have to say continuous blood sugar monitoring could save us the excruciating hassle of being alert all the time. With all that technology, and all the money they ripped of from us, I think we deserve to get a secure treatment. Not a cure maybe, but something that would ensure us that we don't get any serious complications (like blindness per se).

This is all well and good but there are significant safety issues here. Insulin will kill you if you get the dose wrong. Or more worryingly, won't kill you but will kill your child. As a children's diabetes doctor I know this more than most. This article suggests that the only people interested in fixing the closed loop are a bunch of amateur parents - there are large studies going on in several centres across the world looking at exactly this issue - the algorithms to match insulin to glucose are not straight forward: insulin absorbs slowly, with peak action at 2 hours, gone by 4 hours. In children absorption is less predictable - I have a baby on my ward now whose 'insulin-on-board' is 6 hours. The closed loop is the holy grail for T1 diabetes and there is a lot of time, intellect and money being thrown at it. While open source hacking from intelligent and able programmers is always welcome, experimenting on your own children is dangerous and ill-advised.

Why no mention in the article of the actual "bionic pancreas" that has gone through at least one preliminary clinical trial and was published in the New England Journal of Medicine? Seems remiss of the reporter.

http://www.nejm.org/doi/full/10.1056/NEJMoa1314474

A system that aggregated CGMS data from multiple people and used Gaussian Processes to predict blood sugar would likely be the most accurate you could get today.

Kind of weird the dude tried to hide his identity—is he doing anything illegal by treating himself?