• Guest, the forum is undergoing some upgrades and so the usual themes will be unavailable for a few days. In the meantime, you can use the forum like normal. We'd love to know what you think about the forum! Take the 2025 Survey »

Fully Closed Loop Artificial Pancreas? With the latest insulin this is within grasp...

tim2000s

tim2000s

Expert
Retired Moderator
One of the German DIY looping community recently did an n=1 study of using Lyumjev (the new ultra rapid insulin) in AndroidAPS. His results were remarkable. He ended up with 97% time in range (3.9-10mmol/l) and an average glucose level of 6.3mmol/l without bolusing manually or telling the system about any carbs that he'd eaten. I've attached his paper describing it.

Just goes to show that there's hope!
 

Attachments

Interesting article. I sometimes think it going back to pumping and using the variable basal delivery feature with a ketogenic diet could work well for me. But then I am put of by the prospect of dealing with occlusions and haematomas. I am able to achieve satisfactory control doing it manually without difficulty and injections are so easy to do.

The Unannounced Meal mode referred to in the article is a bit of a misnomer. The user still notifies the system an hour before eating so insulin delivery can start increasing in time. Lyumjev and Fiasp may be a bit faster, but it still takes a long time for subcutaneously infused insulin to have an effect. This is a limiting reality, and there is no way around it.

Interesting that they use much longer insulin duration times than manufacturer specifications. It is an important parameter for the algorithms and something pumpers have been questioning for a while. The industry is going to have to adapt to this need for accurate and meaningful data.
 
3.9 to 10 is kind of a big window. I’d bet the limitation variables was cgm and it’s responsiveness. If cgm could eventually sample blood, I bet that window could be more reliably narrowed.
 
Jollymon said:
3.9 to 10 is kind of a big window. ...
Click to expand...
They use the ADA recommendation range for comparison purposes. The ADA recommends keeping HBA1c below 7% and TIR over 70%. With closed-loop it is over 90%, so clearly better control. Even for an anecdotal report.
 
MarkMunday said:
The Unannounced Meal mode referred to in the article is a bit of a misnomer. The user still notifies the system an hour before eating so insulin delivery can start increasing in time. Lyumjev and Fiasp may be a bit faster, but it still takes a long time for subcutaneously infused insulin to have an effect. This is a limiting reality, and there is no way around it.
Click to expand...

Not really. This can be set up automatically (which many people do) such that it is operational within the hour and a half around when you eat meals. This helps to manage meal rises. Whilst subcutaneous insulin is slower than human insulin, if you fully bolus with the meal using Lyumjev, as in the author's case B, it is remarkably effective at reducing the post prandial rise.

MarkMunday said:
Interesting that they use much longer insulin duration times than manufacturer specifications. It is an important parameter for the algorithms and something pumpers have been questioning for a while. The industry is going to have to adapt to this need for accurate and meaningful data.
Click to expand...
Put simply, they don't. The manufacturers provide two pieces of information. Half life and a duration under clamp. The duration under clamp is the observable time of action. Half life indicates the real time in system. It is this latter that creates the long tails that need to be accounted for to reduce hypo risk.
 
Jollymon said:
3.9 to 10 is kind of a big window. I’d bet the limitation variables was cgm and it’s responsiveness. If cgm could eventually sample blood, I bet that window could be more reliably narrowed.
Click to expand...
Not really. As @MarkMunday pointed out, the real limiting factor is the insulin effect and clearance, in terms of responsiveness, and yes, the 3.9-10 is provided to give a comparison aligned with the international consensus on time in range guidelines.

The thing to note is average glucose level and standard deviation, which is ~6.3mmol/l and 23.1%
 
I have higher expectations than 3.9 to 10 mmols. It’s a bench mark, but I want to be better than that. I would hope that someday with better technology (better insulin, and better/faster more real-time test loop) we could achieve something closer to normal.

I’ve always wanted to be normal, so that’s my goal.

I’ve got a flat line on a cgm that regularly says my number range from 3.9 to 6.0. I do this manually. This isn't from a closed loop system. I don’t really trust the cgm- it’s slow, and the numbers vary. A meter can quickly tell me reality. But this is why I think 3.9 to 10 is a big window.

I’m still hoping for the smart insulin assembly, where in the presence of high blood sugar encapsulated insulin will release itself. Then it’s just a matter of us having enough of the encapsulated insulin onboard. I think this would be a game changer- insulin that only goes to work when it’s needed and no need for a test loop. (SIA II)
 
Yet another claim for a closed loop system that avoided testing it at a level of physical activity which cannot be fueled mainly by fat.
 
TypeZero. said:
What’s the benefit of Lyumjev compared to Fiasp? Are they not basically the same speed and duration?
Click to expand...
No. Turns out that Lyumjev is about 10 minutes faster to peak and the tail seems to disappear fully at 5 hours, compared to about 7 for Fiasp.
 
Jollymon said:
I have higher expectations than 3.9 to 10 mmols. It’s a bench mark, but I want to be better than that. I would hope that someday with better technology (better insulin, and better/faster more real-time test loop) we could achieve something closer to normal.

I’ve always wanted to be normal, so that’s my goal.

I’ve got a flat line on a cgm that regularly says my number range from 3.9 to 6.0. I do this manually. This isn't from a closed loop system. I don’t really trust the cgm- it’s slow, and the numbers vary. A meter can quickly tell me reality. But this is why I think 3.9 to 10 is a big window.

I’m still hoping for the smart insulin assembly, where in the presence of high blood sugar encapsulated insulin will release itself. Then it’s just a matter of us having enough of the encapsulated insulin onboard. I think this would be a game changer- insulin that only goes to work when it’s needed and no need for a test loop. (SIA II)
Click to expand...
We use 3.9-10 because it's the agreed standard and allows a reasonable comparison with the metrics provided by commercial offerings. As his paper says, he tends to look at the 70-140 as being more pertinent, and a higher time in that range would be more normal.

Smart insulins would seem to be a sensible approach but as far as I can see, only Novo have gone into any sort of clinical trials on those so far, and they're still at a very early stage.
 
Jollymon said:
I have higher expectations than 3.9 to 10 mmols. It’s a bench mark, but I want to be better than that. I would hope that someday with better technology (better insulin, and better/faster more real-time test loop) we could achieve something closer to normal.

I’ve always wanted to be normal, so that’s my goal.

I’ve got a flat line on a cgm that regularly says my number range from 3.9 to 6.0. I do this manually. This isn't from a closed loop system. I don’t really trust the cgm- it’s slow, and the numbers vary. A meter can quickly tell me reality. But this is why I think 3.9 to 10 is a big window.

I’m still hoping for the smart insulin assembly, where in the presence of high blood sugar encapsulated insulin will release itself. Then it’s just a matter of us having enough of the encapsulated insulin onboard. I think this would be a game changer- insulin that only goes to work when it’s needed and no need for a test loop. (SIA II)
Click to expand...
I have only recently been diagnosed and was rushed to hospital last week with very high numbers, I am recovering now and what I am finding is that I struggle getting any lower than 5.0. It makes me feel very shakey, I think it may just be because I have been running high for months before diagnosis
 
Choosehappy said:
I have only recently been diagnosed and was rushed to hospital last week with very high numbers, I am recovering now and what I am finding is that I struggle getting any lower than 5.0. It makes me feel very shakey, I think it may just be because I have been running high for months before diagnosis
Click to expand...

Diabetes isn’t a short race. It’s even longer than a marathon. It all takes time. Nothing will happen immediately. Pace yourself for this to be a long haul.

If you’re not used to normal numbers then you will feel low.
 
Last edited:
Choosehappy said:
I think it may just be because I have been running high for months before diagnosis
Click to expand...

More thank likely - your body become accustomed to running high (or low) if thats the norm for you.

Slowly you will get used to lower (and healthier) BGs so you'll be ok at 5.

I've been diabetic so long now I don't get good warning signs and can run (sometimes) to 2 or below and feel fine, even though my wife is telling me I'm not (she knows everything....nuff said....)

The way to cope with that is CGM and a watch that buzzes when I'm low - sorted!
 
Back
Top