How far are we from fully closed loop? News from ATTD 2023

One of the great things about the ATTD conference is the updates on the technologies that are upcoming, alongside the “What’s already there”.

It’s very clear that over the past few years, the “Standard of care” in many countries has advanced significantly, with CGM coming as a standard item and Automated Insulin Delivery (AID) taking it’s place as a much more readily available option for treatment.

As an adjunct to that, there is still a significant challenge to people to get hold of many of these systems. India, with now the world’s largest population, is still poorly served in the diabetes technology space, with limited access to CGM and almost nothing on the commercial AID front. And of course there’s access to insulin. All of this is a story in itself, which Jazz Sethi spoke eloquently about in her’s and Partha Kat’s session on disparity. And it’s something that we need to acknowledge.

With that in mind, what I’m about to talk about isn’t even licensed or approved yet, but gives indications on where the technology is going.


Whilst there was a lot of promotion of the new AID systems available on the market, the interesting areas were in the discussions on clinical trials and various presentations. So let’s look at some of those, both big and small.

Let’s start at the smaller end…

Inreda Diabetic is a small Dutch company founded by Robin Koops. It’s the first approved (it has a CE mark) dual hormone system on the planet and currently has around 125 users in the Netherlands.

Robin himself uses it fully closed loop and sees a 93%+ time in range with zero hypos, operating fully closed loop.

It’s not the smallest device on the market, but of course it’s dual hormone, and currently using the glucagon version you get in the orange box. Whilst the glucagon reservoir currently has to be changed daily, Robin mentioned that they’re testing whether they can get it approved for use for longer periods, which would make the process a while lot easier.

Other components currently need regular changing out, but the team are working on a new pump profile and presumably enhancing battery life and other aspects of wear. It’s a really interesting thing to see, given the lack of information from iLet recently.

Diabeloop continue to innovate, with recent announcement of their partnership with EOFlow for the EOPatch pump.

But what’s way more interesting is the launch of their new clinical trial that looks at removing the need for intervention around meals in adolescents, in an insulin only system. From what I understand, they’re confident that they’ll be able to demonstrate some really good results.

Following the outcome of a trial of CamAPS HX, a “full closed loop” algorithm that was trialled in inpatients with type 2, there will be a follow up study with people with type 1, to understand how effective it is with unannounced meals, similar to Diabeloop.

TypeZero Project Liberty

TypeZero (owned by Dexcom) has been working on their next generation closed loop system, and once again, they’re looking at “fully closed loop” or rather, Minimising Mealtime Management.

They’ve undertaken a pilot study in two sites in New Zealand with primary outcomes being to reduce time above 250 and below 54 amongst users who find it very tough to manage their type 1. As the image shows, they’ve been very successful, with huge reductions.

As an adjunct measure, time in clinical consensus range saw an increase from a median of around 37% to around 56%, which if translated to someone with the UK population median, could see them increase to somewhere in the 80% zone. Very much one to watch.


While Medtronic had an innovation suite present, anyone entering had to sign an NDA. I didn’t go in, but I’m assuming they had some new plans for pumps, or sensors, or a combination of both hanging around in there. Given the recent rumours about them putting in an offer for EOFlow, could they have started down a patch pump route and changed their minds?

There was also a presentation from Anirban Roy, of Medtronic, who talked about the use of the recently acquired Klue, a watch app, to detect hand gestures when eating and use that information to determine when to mealtime bolus.

The initial results showed good post prandial glucose levels in the n=17 study, and aligns with what we’ve heard Medtronic speak about elsewhere in relation to how they see closed loops working. There were also discussions about the 780G and its use without announcing meals.

Boris Kovatchev

Of course, it wouldn’t be an ATTD with Boris speaking on something important relating to closed loop systems, and he didn’t fail here either. Boris provided a summary of the various factors that affect a fully closed loop. It breaks down into three areas.

  • Faster insulins – while general insulin pharmacokinetics are nowhere near the glucose absorption curve, Lyumjev presents possibilities. Other companies are working on even faster insulins as well.
  • Mealtime anticipation in algorithms – enabling the system to start to dose insulin ahead of meals based on recognition of when they occur (or as in the Open Source world, scheduling “Eating Soon” periods).
  • Introducing other hormones – testing has been done with:
    • Pramlinatide
    • GLP1 receptor agonists
    • Glucagon (see Inreda previously)
    • SGLT2 inhibitors

All the above showed improvements individually, although, when combined they weren’t always successful.

Another key point from Boris’ talk was the data mining they’d done on Tandem Control-IQ data, discovering that when some users used it full closed loop, they ended up with a higher percentage time in range than when they used it hybrid, however, this is in the context of them having a lower percentage TIR when acting in hybrid, compared to those who had never tried to be closed loop. It gave some food for thought.

The Open Source Community

The open source community talked about the approaches being taken to reduce the onus to bolus, either by eliminating the requirement to carb count and calculate insulin, or to elimate any announcements completely.

Results are promising, with high times in range, even though, again, this is still experimental and n is very low. There have also been a number of lessons learned, which reflects back on some of the comments that were made by others.

Essentially, fully closed loop is not straightforward and comes with challenges.

At the same time as ATTD, a study was published from the Czech Republic on the use of AndroidAPS in a controlled setting, looking at full closed loop versus hybrid. The results showed no difference between the two. The paper can be found here:


I think the key takeaway is that there are still a lot of people working on fully closed loop. Everyone sees the benefit of making this work and the reduction in Diabetes burden that potentially could be obtained, as well as the improvements in clinical outcomes, as shown by the Inreda set up.

The future is very much bright, even if it’s not quite clear what it will look like!


Leave a Reply

Your email address will not be published.