FDA approves Medtronic's 'artificial pancreas' for diabetes | Hacker News Reader

FDA approves Medtronic's 'artificial pancreas' for diabetes

reuters.com

64 pointshelloworld22 comments

Just to give a little more context. This is a start but it's by no means the finish line. Additionally, this is just one company's closed source system that is kind of the "Tesla AutoPilot" for diabetes. It will TRY to keep you in the lines but if there is a real turn or a stop light, you absolutely need to take control.

Go google for "OpenAPS" and "WeAreNotWaiting" for existing open source systems that do more than this Medtronic system does...they are available now and there are over 100 people using them 24/7.

Additional reading here: http://www.hanselman.com/blog/ThePromisingStateOfDiabetesTec...

Source: 25 year Type 1 diabetic who is currently "looping" with an open source "artificial pancreas"

Hey Scott! Been following OpenAPS and WeAreNotWaiting for a while. Just got an old Medtronic pump with the right firmware version. What that community is doing is amazing, and you are a big part of it. Thanks for all you do!

27yo, 7yrs Type 1.

Earnest skepticism: how can you trust Open APS without the rigorous testing required from companies like Medtronic? Sure, FDA regulations are not perfect. But they sure work in a large number of cases. While the DIY spirit of Open APS is commendable, hypoglycemia can be extremely dangerous and the risk is too high without FDA-level rigorous testing.

Here's the inherent problem – insulin is a lot more dangerous on it's own than with any sort of hybrid loop system. The safety having a system like openaps - which includes a LOT of fail safes - is far greater than what a type 1 diabetic can do on their own. 1 in 20 people will die in their sleep because of hypos because he or she simply isn't fortunate enough to feel the blood sugar dipping and wake up to treat. The OpenAps safety measures are far far greater than what we have as a normal person regulating their day-to-day insulin dosing. I will sleep a lot easier knowing that I have something else watching, determining whether the boluses he's given might be corrected slightly so that he doesn't go low or high. Read about it at openaps.org – look at the reference design, and read about the fail safes where the pump will revert back to preprogrammed basal rates if batteries run out. Quite frankly anything with Type 1 diabetes is terrifying. At the end of the day, the risk with type 1 is too high for us to wait any longer.

Because there's always a trade-off between robustness and usefulness, and it's not at all obvious the FDA hits the right balance, especially given the extremely asymmetric incentives they face. If they approve a drug that kills a bunch of people, the repercussions are bad, but if millions die while waiting for a drug to undergo a decade of testing, the FDA isn't blamed.

http://marginalrevolution.com/marginalrevolution/2015/08/is-...

My understanding based on the link in my sibling comment is that the current commercial systems don't inter-operate and that the diabetic must read from one device, calculate a dosage themselves, and then enter that dosage into a second device. At each of the three stages there is the potential for human error and common circumstances such as performing each task in the middle of the night while suffering the effects of sugar imbalance make errors more likely. Not that such circumstances are necessary for errors because the process is repeated by diabetics on the order of a thousand times a year over decades.

My understanding is that the issue with the device upon which OpenAPS relies is that a bad actor could possibly reprogram it over BlueTooth. Doing so would require close physical proximity to the individual. By close to the point that a bad actor could probably physically harm the individual or gain access to the insulin itself...i.e. to within a few feet.

To put it another way, one problem the with FDA approval is that OpenAPS uses a no-longer-approved device and the reasons it is no longer approved are not based on evidence of increased hypoglycemia. Their related to medical device manufacturer liability and hence the cost of insuring against risk.

A few reasons.

* I can read the code. It's not very complex. * There are minimal moving parts. * I can suspend the pump or disconnect at anytime. * It's changing my basal rates (minor adjustments) not randomly bolusing large amounts. This isn't an "overdose" type situation. * I'm fully in control of my facilities. I'm not a child or an older person - each of which are more likely to have hypos. * I have never had a debilitating hypo in 20 years, I'm not planning on having one now. * The system was designed by diabetics, for diabetics. The closed systems that will come out like Medtronics will be VERY conservative. So much so as to be of questionable use to folks who want an A1C <6%

Because of all these reasons, along with 15+ years of pumping alone and doing the calculations in my head (and getting them wrong) I choose to trust these systems.

I found the Hanselminutes Podcast on OpenAPS exceptionally informative: http://hanselminutes.com/514/the-open-artificial-pancreas-sy...

In the 90s minimed brought insulin pumps to market. It was one of the first stock I ever bought and it blew up. They regularly wrote about their primary goal of creating an artificial pancreas. Eventually they were bought by Medtronic and I sold my shares. Glad to see they are finally achieving their long term goal... nearly 20 years later.

It's a great step – looking forward to the options coming in the next couple years with Bigfoot & ILet. Glad to see this on HN – made me finally comment.

We should be complete building my husband's artificial pancreas next week based on open source code. For those interested in what's happening in the community – you might want to take a look at openAPS.

cosign.

Can someone explain why notifying the device of exercise and eating is necessary? Is it because 5 minutes is too low of an interval, or because the glucose sensor is slow, or because the body relies on other eating and exercise cues to regulate insulin.

The sensors aren't perfectly accurate; I don't know about the newer sensors which this closed-loop system uses, but the current generation only fulfills the accuracy requirements by reporting the median of 7 consecutive every-5-minutes measurements and carrying a warning of "don't trust these values if you have reason to think that your blood glucose might be changing rapidly".

My guess is that providing notifications of exercise and food (aka. "I expect my blood glucose to go (down|up) right now") allows the model to be tuned to respond more aggressively to new data at some times and less aggressively at others.

(The "less aggressively" side is probably most important: If the sensor suddenly reports a dramatic change in blood glucose in the middle of the night, the system should probably respond with "HEY WAKE UP THE SENSOR IS BROKEN" rather than "let's kill the patient".)

Exactly. Not to mention they are testing interstitial fluid and not whole blood, which introduces significant delays.

I've found that the interstitial delay is pretty minimal; maybe 5-10 minutes behind fingerpricks (which themselves lag somewhat behind arterial blood, especially if your hands are cold). The dominating factor is the 15-20 minutes it takes before the median of a window reflects the new data.

Sure, but we agree every minute counts. Also, CGMs in arms differ from belly, etc. 10 min when insulin starts in 30 is a non-trivial amount of time. Point is, we def need faster insulin and faster, more accurate CGMs.

Because if you start exercising or eating NOW the insulin administered won't start working for 30 min and won't peak for 90-120 minutes. Because food hits you fast and by the time the system notices you'll have already drifted "out of the lanes."

Diabetes Closed Loops are the same problem as controlling the Mars Rover. You move the joystick and nothing happens for x minutes. Then you wait another x minutes to observe the results.

A couple things - the insulin currently on the market works too slowly for a pump right now to fully close the loop – this functionality – notifying the pump of exercise and eating is part of the user control needed in a "hybrid closed loop" such as this. The user's insulin sensitivity given before, during or after exercise are very different than at other times. Different types of exercise affects body differently as well –HIIT for example raises blood glucose for a couple hours during and after – and then will significantly increase the sensitivity to insulin and many type 1's will drop rapidly. Aerobic activity will tend to drop blood glucose rapidly on it's own. Notifying the pump allows you to drop the levels given – something users do on their own now.

In people without diabetes, the beta cells make and release some insulin all of the time. This is a “basal” level of insulin that helps the body to function. When a person eats and their blood glucose level rises, the body releases an extra burst or ‘bolus” of insulin to help bring the blood glucose back into the target range. The insulin taken to manage diabetes is designed to work the way the body works as closely as possible. http://www.idf.org/worlddiabetesday/toolkit/pwd/insulin

I've always been fascinated with metabolic diseases/syndromes like diabetes. It seems to be a symptom of the first world, i.e. affluenza, but there's this weird sort of trade-off where it can manifest itself (type-2 specifically) and become aggressive at a moment's notice.

Clearly there is a strong genetic component, but possibly a strong environmental component as well. I think the markers we have right now (A1C and blood glucose levels) are insufficient in understanding the underlying mechanism of action for the disease.

What's even more interesting to me is the relationship between endogenous growth hormone and insulin receptors. The relationship between IGF-1 and insulin in particular is very interesting - IGF-1 binds to insulin receptors.

I hope that as we continue to advance in cybernetics and genomics (artificial organs are the first step to machine/human hybrids), we will gain a better understanding of the underlying mechanisms of the common ailments in our society.

You don't know what you're talking about at all. Read what type 1 and type 2 diabetes are before you talk about them.

Clearly, I know the difference between type 1 and type 2. Type 1 is primarily genetic, which I may not have explicitly mentioned in my comment, and type 2 is a mix of genetic and environmental factors.

Thanks for the two sentence reply fam, it really elevated the discussion.

Reply on news.ycombinator.com