Initial Loop review

2 months on Loop today. Initial review is as follows (this may change down the line). I’m sure you all will provide feedback if your assessments differ:

I recently did some fine tuning on baseline insulin delivery rates and and started running the system in open loop. I’m not seeing a ton of difference in basal system behavior. During some time periods, open loop behavior looks better than closed loop, absent any influence of exercise variables. Its not a medically significant difference.

For sure, system controls are better than pre-loop. This is mostly due to insulin delivery at mealtime. That is MUCH more effective. System variability has decreased significantly because of this.

A1c, was identical to pre-loop (I’ve only been running loop for 2 months, so this is only 2/3 Loop data, and 1/3 non-loop data). a1c = 6.1. So, the average stayed the same, but system variability definitely decreased. That’s real clear.

The main benefit from system automation is that it adjusts the basals when there are impending lows and when basal dosages drift over time. But, it seems more significant that insulin delivery surrounding meal times is producing much, much better results. I don’t think I can really attribute the majority of my improvement to ‘system automation.’ Although, it does soften the lows, which is invaluable when its needed.

One thing that got me is that as the basals drifted over the past 2 months (I gained 10 lbs), I didn’t notice because the system was overcoming it. So, when I opened up the loop, things were more out of whack than I expected. Not terrible. But, its good to periodically run open loop, just to prepare for a Loop failure due to lack of sensor supplies or something.

My BT communications are pretty unreliable with the G6, compared with the G4, so that has been the most common point of failure. There are concerns about the newest iPhone update. I have not updated. They are still working on it. If you are considering starting on Loop, you might wait for the all clear that its worked out.


Thanks @mohe0001, I find this kind of personal-experience log with new tech really illuminating. Question for those of us not up on the nomenclature for this system: can you explain what you mean by open loop versus closed loop?

If I’m not being too forward, I will try to explain:

When using the Loop app, you have the option to open or close the Loop. When the Loop is closed, all the devices are communicating with each other as intended - Dexcom’s data is used in conjunction with your settings and carb inputs to predict bg levels and calculate changes to basals based on those predictions, and finally communicate those basal changes to the pump.

You can opt to open the Loop which stops that communication and basal adjustment. The calculations are still performed, but the basal is not changed in the pump. You may opt to do this when you are adjusting settings or you know that Loop will not be effective the way you want it to. Caleb will open Loop if he has a stubborn high that isn’t responding to insulin because he does not want basal to be shut off which will only compound the issue. This happens sometimes on the third day of a Pod, for example.

1 Like

Ah, I see. So “open” is analogous to Manual Mode in the 670G, which is what I was wondering, though I’m not sure how exact the correspondence is. For one, I don’t know whether the automode algorithm keeps running calculations when not engaged. It might, but I don’t recall it being discussed in those terms. For another: the 670 reverts to your original basal profile settings when you take it out of automode. Is that what Loop does as well?

On the whole it sounds like it’s simpler to engage/disengage (or close/open) Loop. It’s kind of a pain with the 670. Easy to drop out of auto, but re-enabling it requires you to go through a set of steps to ensure you meet certain conditions. Calibrations and whatnot—hazy on the deets as it’s been a while. But if you’ve been bumped out for some reason, which happens fairly often, and inevitably in the small hours of the night, you can’t just switch it back on and go back to sleep.

1 Like

Why did Medtronic decide to make auto mode so finicky? If you’re in auto mode, the pump should stay that way unless the user decides to change it.


I think pretty much everything that made me give up on automode can be explained by the need to make the system super-cautious to alleviate the concerns of the FDA. We get used to dealing with insulin because we’re on it 24/7/365 for years, decades, a lifetime, but the mentality of the medical profession has always been that you don’t just let people go around making up their own doses of potentially dangerous injectable drugs. I don’t know if there are any other diseases that require them to stand back and leave day to day decisions like that in the patients’ hands. It goes very much against the grain, and that goes double for the FDA. Unfortunately there’s no other way to do it, so they have to accept it. And all of that goes triple when you’re talking about an untried (at least in the commercial realm) automated delivery system. So it’s very locked down with all kinds of security doors in place. If it sees something it doesn’t like—often a sensor reading that doesn’t make sense to the algorithm, also there’s a limit on how long it will let you go without a delivery if you’re running toward the low end of your range and don’t need one, which is infuriating to a lot of users—it bumps you out and alarms you awake, and then you have to test, calibrate, whatever is required to convince it that there isn’t a problem.

A lot of that makes sense. From what I’ve read about Tandem’s loop algorithm, the pump will not kick you out of it at any point. It’s either enabled or disabled, the differences in the two algorithms between Medtronic and Tandem is night and day. With the Tandem algorithm, simple is the word most often used to describe it.

Tandem is getting the benefit of not being the first to market with this technology. Not to mention interoperability with a superior CGM. Reports are that Medt is looking at improvements, including working with Tidepool and opening up to Dexcom integration. Hope it pans out.

1 Like

I’m not impressed with the data that Tandem submitted to the FDA for their control-iq. I expected the group using control-iq to have a much higher time in range than 70%.

The Medtronic 780 is around the corner, it looks promising so far. Let’s keep our fingers crossed for Dexcom integration.

1 Like

Yep, ‘open’ literally means Open Loop System, like a traditional insulin pump where we deliver commands. And, alternatively, a "closed loop system,’ where the system will increase/decrease basal without user input. Closed Loop system operate autonomously, like a refrigerator that maintains constant temperature if your kid keeps opening the door. You don’t tell it to lower its temperature, it just does that all on its own.

Loop also reverts to pre-programmed basal settings in its ‘open setting’ or manual mode. What surprised me was that I still had far better system behavior (because meals came out so good). I think thats been the most significant benefit, for me. Its easy to open/close loop - you just select it.

One of the things, @DrBB, that you mention - I did find it really surprising to see how often Loop suspends my basal insulin delivery. I’m pretty insulin aggressive on the whole. But, sometimes, for hours it delivers nothing. Strange. Wouldn’t expect that was ever needed. It works, though. It will bump me back up into control limits. I never would suspend insulin delivery for that long on my own.

See where it suspends basal delivery for three hours here? Bottom plot. Wowzers. @DrBB, it does this will some frequency (especially when I over correct).


What is that report?

That’s a Tidepool daily report. It also will reveal the underlying values if you hover the mouse pointer over the graphic, like any portion of the basal profile.



You can watch my data here, if you want:

1 Like

And if anyone wants to watch mine, it’s at . Just past the 3 month mark and am loving it so far. Really cut down on my lows and not spiking as much as before. SD is also down a bit.

Just had my A1C drawn a few days ago and waiting on the lab results.


Everything has been working soooo well.
But, I’m not sure whether to expect much performance now that its getting cold.

Its been 45 degrees at night. If I use a blanket, then I loose closed loop communications. I want a blankey. Last night I was freezing (from a pretty low BG). I really piled on the blankets. It didn’t autocorrect all night. Any ideas?

Frost comes this weekend.

With my Med-T pump, the weak link in the radio communications is between the pump and the Riley Link. If that’s true for the Omnipod Loop, where do you place the Riley Link while sleeping?

With my pump I use the generous infusion set tubing length to place my pump to my side, under the covers, with the RL on top of the blankets above it while it recharges. Since I’m a side-sleeper, I can move from left side to right side without interfering with the short distance the RL must jump to talk to the pump. I rarely experience data drop outs while sleeping or even other parts of the day.

If the Omnipod Loop shares this communication weakness, it might be worth experimenting with reducing the distance between the pod and the RL. Seems like you could use an ace-bandage to wrap the RL right next to the pod if you’re using an arm placement or otherwise positioning the RL so that it has a clear and short distance to bridge to the pod.

Perhaps placing the RL above you somehow so that it can “see” the pod in your preferred sleep positions. Some beds, for instance, feature a shelf nearby.

I don’t think that blankets are the impediment here. More likely it is distance and especially placing your body between the pod and the RL as you shift sleep positions.

1 Like

Looking back at my A1c performance after I started Loop, my A1c actually went up 0.2% at the three-month mark and then dropped 0.7% at the six-month draw. Some of that could be A1c measurement noise variation but you get the idea.

I hope you get good A1c confirmation but your better every day glucose performance will eventually make its presence felt in the long run. Monitoring time in range, % hypo, and glucose variability are stronger performance measures from my perspective. But I know that the A1c is the one indicator the doctor will notice.

1 Like

Would you share your references on the Tandem Control-IQ algorithm? I’m anxious to learn more about it before I take the leap and upgrade. Basal-IQ is giving me what I need at the low end, and if Control-IQ won’t match or improve that algorithm I’m not sure I want to upgrade. The Tandem rep has been unwilling to share anything about Control-IQ because it’s pre-market.

I have asked for the opportunity to test it on a loaner pump for a month or so before upgrade. She said that’s unprecedented but to ask again when it’s officially launched. The problems people report about 670G auto mode makes me take pause. From what I understand from the rep and elsewhere, once you upgrade there’s no going back.

I was just gonna ask the same. I dont know much about the Tandem algorithm.

@Paytone, you have any resources on the current algorithm? Can you say about your current experiences? I know nothing. BTW, thats not unprecedented. Many of us have had loaner pumps before. I wonder why she said that.