At three years, Loop shines but not perfect

Today marks three years since I first started Loop, the do it yourself automated insulin dosing system. It has improved my glucose management by giving me better results with less effort.

My CGM time in range increased, often to a rarified level of 90%+ and I set my range limits to 65-140 mg/dL. Serious hypos, those < 55 mg/dL are limited to less than a few minutes, on average, each day. Standard Deviation as a proxy for glucose variability is often under 20 mg/dL and rarely over 30 mg/dL.

My case is not typical

Now, I know my situation does not represent the average person with T1D. I’ve been using a cab limited way of eating (<70 grams/day) to tame glucose extremes since 2012. Starting one year ago, I further restricted any foods containing grains or any added sugars. I did this in response to a coronary artery disease diagnosis. I don’t actually count my carbs each day but I’m easily under 30 grams/day.

I feel certain that Loop would not perform as well for someone who wanted to consume a diet of highly processed carbs on a regular basis.

Changing basal insulin more powerful than it appears

Loop performance is subject to how well the user selects appropriate basal profiles, a daily schedule of Insulin sensitivity factors, and insulin to carb ratios for meals. Honoring good pre-bolusing hygiene is essential to reasonable post-meal traces, even with Loop. Subcutaneous insulin injection will never compare well to a non-diabetic’s pancreas release of insulin directly into the portal vein of the liver. Loop can’t change that fact.

Back in the '90s, when I first read of the concept of a closed-loop insulin dosing system, I was highly skeptical that modulating basal insulin delivered by an infusion pump could ever catch up with nutrition metabolism. This is one of the things that really surprised me about Loop.

An amazing system that I like to watch in action

With Loop settings optimized, I never get tired watching Loop slow, stop, and then reverse a rising or falling glucose line to return it to the target I set. It can affect positive change within about 15 minutes, an amazing feat considering the sub-q insulin delivery route.

Unlike commercial systems, Loop permits setting your glucose target to whatever you want. I set mine at 83-84 mg/dL as I consider that a normal and optimal glucose level.

Loop is not perfect, neither am I

My metabolic life is not perfect, however, with Loop. A recent incident with an errant glucose reading from a newly inserted sensor was made worse by me making a rookie mistake. Seems Loop can’t fix my human shortcomings!

Here’s what happened. I put a new sensor into service and like many sensors in the first 24 hours, this one was not ready for prime time. In fact, it was much worse than the average first day sensor errors.

It first dove for the hypo range showing over a two-hour period persistent lows in the low '40s. I was fingersticking, holding back my desire to over-calibrate, and knew that my actual blood glucose was in the mid to low-60s.

Then this ill-behaved new sensor decided to move in the other direction. Now, when sensors over-react to the low side, Loop just shuts off basal delivery. Its response to a low-60s trend is the same as a low 40s trend – it can’t step on the brakes any harder than a zero basal rate.

On the high side, however, Loop can continue to add more basal insulin subject to user-adjustable delivery limits. I am fairly liberal setting these limits and this didn’t help the situation. The false sensor glucose topped off at 274 but a fingerstick only showed 136.

Humility still applies

My mistake was not paying close enough attention to what was actually happening. Loop was over-delivering basal insulin and I should have stopped it. This is a rookie mistake and a reminder that even with an automated system like Loop, you always have to keep your head in the game.

Before I knew it, my insulin on board number had ramped up to 5 units and my blood sugar was not at 274 but 136. I had to eat several glucose tabs to compensate. Here’s the graph of my mistake.


CGM accuracy can be Achilles’ heel

I’ve read criticisms here that automated insulin delivery systems are not reliable due to the underlying inaccuracy of glucose sensors. The case I show above illustrates well that weakness. But this situation for me is an outlier. This set of circumstances, including my inattention, doesn’t happen even once per year. And my safety can be improved with a lower delivery limit.

In summary, I love my Loop system. It has improved my glucose management and my quality of life. Almost every morning, I wake up in the 70-99 range. These are early days of automated insulin dosing and I think the algorithms will get better. I also think the user interfaces will do a better job of balancing the safety settings without inducing alarm fatigue.

Of course nothing will work as well as the glucose system that mother nature produces but without the mythic cure-in-five-years on the horizon, I will continue to happily live with my Loop.

Past Loop posts

Original 2106




Nice summary, and congrats on your success.


Thats a great summary of three years, Terry4.

The big difference between you and I is that becuase of your gasteoparesis, I bet the basal automation is a big deal for you. For me (no gasteo), the biggest improvement has come in the form of more effective meal bolus. I am still new and figuring out meal dosing, but insulin delivery at meal times is MUCH more effective with loop. I can keep under 180 post-pranial with high carb meals. That was not possible for me before.

Loop delivers meal insulin more effectively. Because of that, I run much better even just using open loop settings with the system. Its a thing. But, dosing isn’t straight forward. I’m moving away from carb counting. I just have to know how many units each type of meal is. I’m returning to that old way of doing things, with some adjustments.

1 Like

Thanks for the update/summary! Really impressed with what you’ve achieved with diy looping. Might push me over the fence from mdi to pumping

1 Like

I don’t sense that my gastroparesis plays much of a role in my metabolism. While I do sense mild symptoms, they interfere little with my day to day glucose management. Good to read that Loop helps you with meal bolusing. I think that a dispassionate computer is better overall in the minute to minute decision-making. I’m glad to read that Loop is helping you. Good luck going forward.

I know that MDI can work well since so many people successfully use it. I had a good MDI experience when I took a several month pump vacation a few years back. I used Tresiba and Apidra pens combined with Afrezza mostly for corrections. It worked pretty darned well but I soo n transitioned to Loop and have not looked back. It’s good to know that I can use MDI as a backup if and when needed.

While I think the human brain excels at making overall strategic decisions based on creative insights, I concede that Loop is better at a tactical level and never sleeps, never tires.

I hope that automated systems can improve quickly enough to take over for me as I age. I think I’ll eventually need a system that’ll only require infusion set and glucose sensor maintenance. In other words, as my brain becomes less capable to make the many decisions I currently make, it would help me tremendously to off-load that cognitive duty to an algorithm.

I think people who are successful using MDI could also use that kind of help as they age. Don’t get me wrong, I can hope that my brain and cognitive abilities remain as sharp as they are today until the day I die, but I can see a trend in me that forces accepting a certain reality.


Excellent synopsis, thank you and congratulations!

Looping seems to be giving more people more time away from having to manage the lack of endogenous insulin. It complements all the other dynamic mechanisms going on in our bodies that defy making insulin-dependent diabetes a simple math problem.

I was recently at a multi-day meeting with a group of D-Advocates, most were either on the T-Slim Basal IQ or were looping. Compared to last year, there were almost no discussions about BG excursions, or “how many carbs in this dinner roll?” Instead, we were able to focus on the task at hand, advocating for PWDs. It was a very freeing experience.

Just prior to this meeting, I had my Medicare-mandated 90-day endo appointment, where I asked about moving to the T-Slim Control IQ system. My endo’s response was, “you’re so well-managed that I don’t see where you would significantly benefit.”

My response was, “Thanks for the attaboy, but it would be nice to let go of the controls a little bit.” Then I sighed in my mind.

Like you, I do have concern about aging and having to let someone else manage the minute by minute, day by day details. Effective looping would be huge.


Thanks for the periodic updates. It’s great to see your closed loop history. I’m patiently waiting for Control IQ to just take some of the daily tasks off my hands. Specifically hoping it helps my overnight control.

Thanks agian.


While I consider my carb-limited way of eating and Loop as two of my most effective assets to manage diabetes well, I consider a healthy cognitive function my best asset, by far. Anything that diminishes my brain function will adversely affect my diabetes management. It’s my biggest fear as I move into my older years.

Off-loading some of that responsibility onto technology is one way I hope it plays out – soon!

Almost every person’s first impression of looping cites better overnight control (read: better sleep) as a large benefit. I suspect Control-IQ will do a good job with this. Tandem’s got six weeks to get this system introduced before the 4th quarter 2019 ends. Enough with the suspense!

1 Like

Great summary, thanks for the post! I wonder if loop would be even more effective with Fiasp? Since switching to Fiasp, I no longer have to pre-bolus by much and I often see trends turn around within 15 minutes after doing a small correction (as long as my blood sugar isn’t too high to start).

I am hoping that I will be able to loop, either with a do-it-yourself system or a commercially-available system, within the next couple of years.

My biggest concern is my recent discovery that the Libre is not accurate for me. It consistently reads 1-2 mmol/L lower than my actual glucose number, which has resulted in an unexpected rise in my A1c since switching over. I’m doing more fingerstick to compensate, but obviously for a loop system, that would not work.

So, it may come down to deciding whether I want to take my chances with using the inaccurate Libre for free with a do-it-yourself loop or pay for Dexcom out of pocket with either a do-it-yourself or commercially-available loop system.

In theory, I think Fiasp would work well with Loop. What undermines that belief are the many accounts of people starting Fiasp with great success and after some time their glucose control becomes worse than before. Some people have experimented with diluting Fiasp with Novolog (close formulations) with some success.

Here is one Fiasp account written by Tim Street from the UK.

As I recall, he wrote many pieces about his experience with Fiasp.

Yeah, you do need a fairly reliable CGM data stream to reap benefits from Loop. I think the system can tolerate inaccuracy in the range of +/- 10% but also needs to be reasonably precise or consistent.

In your case, I think using an inaccurate Libre is probably worse than foregoing the hybrid closed loop altogether.

By the way, have you experimented with an aggressive hydration program as well as trying various body locations to improve the Libre results?

1 Like

Interesting, I hadn’t heard of this before. I had heard of infusion site irritation. I’ve been using Fiasp for almost three years with no loss of efficiency.

I have not done much experimentation yet. I’ve only been using the Libre since July and only discovered a week or two ago that it was inaccurate. (I’d expected my A1c to be lower or the same as before, and it came back 0.6% higher, so I began testing much more frequently and discovered the inaccuracies, except for the odd time it’s accurate, mostly during lows.) I’ve only worn Libre sensors on my arm. Recently, I’ve lost two sensors after just five days, so I’ve added some tape over my current sensor to secure it better. I do have a MiaoMiao (first generation) and am hoping that it may be more accurate with calibrations, but right now I mostly use it at night. If I were to use a do-it-yourself loop, I’d have to use MiaoMiao anyway to get continuous readings with the Libre.


Thanks for posting this Terry. I’ve experienced a rash of hyperglycemic episodes recently which I’d thought were caused by coincidentally bad infusion sets & sites. Now I’m wondering if it’s the FiASP (pumping it 8 months now).

Next site change I’m going to try Humalog for comparison.


I had super initial results with Fiasp, then they did diminish.
I went back to Novolog for a short period.
I realized that Fiasp was still FAR better than Novolog, and went back to Fiasp as soon as I could!

I also find that Fiasp works better for me when using a longer cannula. I have to use a 9mm cannula or I have a lot of other issues, like having them leak out, not absorb well, site irritations.

1 Like

Three years?! Already?! That means I have been thinking and dreaming about this system for three years?! Wow, it has been a long time. And I remember the frustration in trying to find a magic pump. And watching your posts and talking with you about the system, and seeing others at conferences using the system. It just amazes me how great these systems can be.

And your outlier of a sensor not being accurate is always a problem. But I have been very, very fortunate that my current system runs pretty darn accurate. And I must say when they first talked about these systems, I was super concerned about the accuracy and if it was off and it sent me crashing. But using the Tandem IQ, which turns off when heading low, I have had no problems. Even on the times when I do test (clinical trials) and it is off, if I wait 15-30 minutes, it usually catches up.

The other thing that makes me so darn proud of all of us, we try and manage the blood sugars to the best of our ability and we usually miss. If I look at how many times my pump turns on and off during a 24 hour span, I think no wonder I was a “brittle” diabetic in my early years. Who could possibly make that many changes every darn day.

It amazes me how many new pumps are coming down the pipe line!i can’t wait for this new technology. I know many are not sure about turning over the control but I will say it has been liberating! And yes, I know I am still the person responsible for all this and when in doubt, I check. But checking the system once a month or once every two months or six months, so much better than 8-12 a day!

@Terry4, your results are amazing and I love that you don’t make it roses and sunshine. It is work! I always tell people pumps can make your life easier is some regards but they do add a lot of work. But with all this technology, it is getting easier! Keep it up! And while me numbers are no where near yours, I am very happy with where I am at and am hoping to get a complete system one day so I can give you a run for your money!
But I love pizza and Mexican food too much!


I look forward to the day when technology can move to the next level of “smart” – when it can learn my personal quirks and make future adjustments using that learning. In the meantime, I’ll do everything I can to extend the time that my cognitive function can continue its ability.

From your comments and others about Tandem’s Basal-IQ, I think I would be using it if not for Loop. I look forward to reports when the Control-IQ debuts – soon, I hope!

1 Like

My daughter consumes 150 to 200g of carbs a day. Loop is performing much better with the new carb model in Dev! We are doing about as well as Pre-loop now and I don’t find myself arguing with Loop as much anymore.

1 Like

For those who may not be familiar with the terms, can you describe a little more? From reading some on the Looped Facebook group, I’m aware of the hundreds of people who more recently adopted Loop using Omnipods.

I know that Loop has various “branches” or versions and I think that development is one of them. Can you tells us what this new “carb model” is doing for you and your daughter? How does it work? How is it different than the current branch’s carb modeling that is not in development?

I should spend more time reading and understanding more about what is going on “under the hood” program-wise in Loop. Learning from others like you helps me and I’m sure a few others here.

That “smart technology” is in the works. This is one of those things that makes me so excited about the Beta Bonics pump that is in trials. I remember watching a video where they talked about no carb counting! Could you imagine, no carb counting?! The pumps is just told if it is a small, regular or large meal and it starts to learn what that means for the person wearing the pump. So my 140 grams of carbs on my pizza night would be a large and it would learn over time how to dose for that! One less thing to think about!

Nice report! I have always resisted pumps because I struggle with the idea of attached devices. Oh, and cost. I am in Canada and the cost of the Omnipod, the only pump I would consider, would pretty much break the bank. On low carb (like you, less than 30g pr day), and MDI my stats are very similar too yours. My last A1C was 5.1, and my SD is approx 22 on average. I have a tighter range (70-120), but even so your time in range is definitely better than mine. I keep being told to do loop, and I am going to discuss with my endo and see if there is anyway I can get some monetary support. Do you use Dexcom 6? Do you know if it is possible to do Loop with a Freestyle Libre / MiaoMiao combo?

I currently use a Dexcom G4 CGM, will eventually transition to the G5 and then the G6 over the next year or so.

From Loopdocs:

Can I use Libre sensors with a reader like Miao Miao?

If you want to use a Libre sensor you will need to modify Loop to accomplish that use. Loop code does not natively support that sensor, nor readers.

With the current excellent glucose management you enjoy, I think the best thing that a system like Loop could offer you is similar performance with less effort. For me, it also offered better overnight performance. If you are satisfied with both the performance and effort involved with your current system, I could understand your reluctance to change anything.