Close-It clinical trial

I’ve mentioned previously a clinical trial I was involved in running, in which 20 people used AndroidAPS and research YpsoPumps.

The next trial is now up and running. It’s called “CLOSE IT” (or sometimes CloseIt or Close-It) which stands for “Closed Loop Open SourcE In Type 1 diabetes”. No in case you’re wondering, I didn’t come up with the name.

The top-level description is:

Randomised controlled trial investigating the safety and efficacy of an open source automated insulin delivery system without manual mealtime boluses, in people with type 1 diabetes.

Overview

Basically, it’s using Lotus (which is research-oriented software that’s descended from AndroidAPS) with Dexcom G6 CGM and the latest version of the research YpsoPumps. The Automated Insulin Delivery (AID) algorithm being tested is oref1, the same as used in AndroidAPS, OpenAPS, etc.

This time there will be 75 participants (spread between a site in Australia and one in New Zealand). Each person will be using the system for at least 6 months.

In the first 3 months they use it in “hybrid closed loop” mode, where they announce all food, bolus for meals, etc. Similar to most commercial AID systems really.
But in the second 3 months they’re randomised so that half of the people will keep operating the same way, and half of them will move to not announcing food or giving manual boluses.
There’s a 1-month smaller extension trial which will be comparing the results of using different insulins (NovoRapid vs Fiasp).

This is an RCT investigating the behaviour of the “fully closed loop” system and lifestyle I have been using for some years now. Instead of just individuals like me documenting our experiences, it’s intended to provide robust data which will stand up to academic scrutiny. The recruitment is quite varied: some participants will have never used an insulin pump before, some will be seasoned pumpers, some will have used commercial AID systems. Similarly there will be varied CGM experience levels. Unfortunately for some of my audience, people who have been using OS-AID systems (e.g. Loop and AndroidAPS) are excluded due to the design of the trial protocol.

It won’t surprise you to learn that one of my roles is as part of the team providing the technical infrastructure for the system. I’m using Lotus every moment of every day myself, although I personally am using an experimental branch distinct from the stable version that was set for the trial.

Timeframes

Of course those 75 people are not all running together but will be staggered in time, so the last of the cohort won’t be finished before 2024. Unfortunately clinical trials take time when you dot all the i’s and cross all the t’s.

Last month a local TV station ran a story on the trial.

As expected, some of the technical details did get lost in translation, but it’s not a terrible story. I did find it amusing to see the sample name they used on their made-up graphs.

Equipment

The YpsoPumps used in the trial are a later version than the “OPN” version used in the previous trial. They’re similar to the public “Dose” (1.5X) YpsoPumps used with CamAPS, except that the public pumps use an extra encryption layer that we cannot currently access. So while we’re able to show what these pumps can do, we can’t yet use our software with the public pumps.

The Android devices being used are Cubot KingKong Mini 2 Pros that are locked down and administered using Google infrastructure.

Lotus itself is not currently available outside clinical trials, although much of its code will be accessible in the coming months. We’re working on that! In the meantime, similar functionality is used by thousands of AndroidAPS users around the world every day.