Hacker News new | ask | show | jobs
by spaethnl 1585 days ago
I think this article does a great job covering many of the difficulties of T1D.

One component I think was under-emphasized is the fact that correction insulin doses are not based on what your current blood glucose(BG) levels are, but on where you predict they will be when the dose really starts taking effect.

Take for example a current best case scenario of having a Loop system via a continuous glucose monitor (CGM) and pump:

If you took a reasonable guess dose for a meal then check your BG levels after the meal, you may find that you have a steeply inclining graph. Here are two possible cases:

  A. You took a correct dose and the timings are slightly out of sync, but BG will eventually turn around.

  B. You under-dosed and will need to either take a correction dose, or wait a long time for the basal dose to fix it.
It can at times be very difficult to distinguish between A and B, and guessing wrong has consequences. Futhermore, you won't really know which is the case until sometime later.

If you are wrong about A then you did nothing, but really you needed to take an correction dose. You won't find out you were wrong for a while, in the meantime your BG is sky-rocketing.

If you are wrong about B: then you over-dosed and are running low. How much did you over-dose? How many carbs should you consume to correct?

Because your CGM only updates every 5 minutes, and typical rapid acting insulin takes about 20 minutes to really get going, this cycle can play out every 25 minutes or so until you have stabilized your BG. all while you may have unhealthy BG levels, and you may be Yo-yo-ing.

This is very slightly mitigated by using an ultra-rapid insulin like Lyumjev, or Fiasp, which can get going in 15 minutes, giving you a tighter loop.

It would be very helpful if:

  1. ... CGM devices had options for more frequent updates during highs and lows. Tighter feedback loops could go a long way.

  2. ... pumps could dose insulin and glucagon automatically.

  3. ... there were even faster acting insulins. This is tough because most insulins are injected interstitially, which takes time for your body to absorb. Maybe an out-patient implantable pump that could inject intravenously would help?

  4. ... there were BETTER INSULIN PUMP SOFTWARE for calculating doses. I have a Tandem T:Slim x2. I  can tell it how many carbs I am eating. Only that. It doesn't count or learn from: proteins, fats, what kinds of carbs, or what specific ingredients are there, or their ratios. All of this can dramatically effect how quickly your BG rises, bringing you back to the original problem of guessing. It should be possible to select from a database of commercially available food and manually provided recipes.
4 comments

Re: 3, such technology did exist and was commercialized in the early 2000s, and was well liked by those who got to use it (https://www.researchgate.net/publication/38063312_The_Implan..., but it had a host of issues that made it difficult to be widely viable, and ultimately it was discontinued.
I mentioned it in another comment, but inhaled insulin does act much faster than injected insulin. Of course, one downside is that it requires manual dosing; it can't be managed by a pump. But it's extremely fast (and finishes quickly too), so you can get much closed to dosing based on current numbers (obviously you do need to anticipate a tiny bit, but it's drastically less).
Yep - my 10 year old son has a x2 pump and a Dexcom G6. Hopefully the software starts to improve at a faster pace with smartphones becoming more tightly integrated:

https://www.tandemdiabetes.com/landing-pages/remote-bolus

Yeah, I have been looking forward to that for a while. I'm hoping that if their app isn't flexible enough to do menu-based dosing that it will be possible to side-load my own app to do that dosing.

As an aside, the OmniPod 5 was just released and already has App control[1]. The tubeless aspect of that is very attractive.

[1] https://www.healthline.com/diabetesmine/omnipod-5-tubeless-s...

Dexcom being able to send more frequent updates during a hypo would be nice, but not sure if it would offer anything other than noise. We just finger prick during hypos, because I want more immediate information.