Now that many commercially available automated insulin delivery (AID) systems are available worldwide, is there still a desire for open-source AID (OS-AID) systems? The short answer is yes, but what does that entail? I interviewed Dr. Rayhan Lal, a dual-trained adult and pediatric endocrinologist at Stanford University and a PWD, to understand what is available.
Some background
A couple of months ago, Dr. Lal suggested I start a GLP1 agonist to help reduce my lifetime renal and cardiovascular risks and slow down carb absorption. However, he was worried about starting one while using the iLet due to the inability to make setting adjustments. So, he volunteered to help set me up on one of the OS-AID systems. This would require my iPhone, Dexcom CGM, Omnipod DASH pods, and training over Zoom.
What I didn’t mention is that Rayhan was trained as an electrical engineer before pursuing his career in medicine, so his level of understanding of this topic is very high and more in-depth than most people. He provided me with some straightforward and minimal instructions on what to do before the zoom, which I did, and only had to ask a couple of questions – very much not an electrical engineer myself. Rayhan created a Nightscout website for me, and I got to choose the name for my APS. I lovingly chose “Pancreassucks.”
Dr. Lal then spent 90 minutes on Zoom walking me through the setup in the app, going through settings recommendations based on some data – total daily dose (TDD), average daily carb intake, and BMI. I am always excited to try something new, but this algorithm seemed particularly intriguing as it does frequent “micro-boluses” to bring glucose levels to target. Designed by PWDs, the system helps reduce the work required to achieve glycemic goals.
Wait, what system am I using?
I went to clinic at Cincinnati Children’s so excited to tell people about “iAPS” but quickly realized I was not even sure what system I was using. Was it loop? No, but that is another OS-AID algorithm running on iOS. I had asked Rayhan many questions during our Zoom, but most were practical and detail-oriented versus big-picture. It didn’t even occur to me to ask exactly what the algorithm and app I’m using were. I trust Rayhan, and this is the algorithm he was using, and that was enough for me.
But I wanted to be able to answer people’s questions, so I started texting with him and quickly realized I had to have an actual conversation to figure out what the system was, how he chose to use it, and recommend it to others. So, Rayhan generously had another call with me, during which I asked some questions and mostly listened to how the system, which I learned was called iAPS, was developed. Here’s a bit of my “interview” with Rayhan.
Marissa: What exactly is the iAPS system and how was it created?
Rayhan: Before we had commercial systems, the first widely available closed-loop algorithm was OpenAPS. For a couple of years, it was implemented as code that ran on a system on a chip that you’d put in some container (e.g. Altoid tin, 3d-printed case). This would modulate insulin delivery in the background without much user interface. Since people were already carrying microprocessors in their pocket it became clear that these algorithms should run on your phone. Loop, a different control algorithm, was developed into an iPhone app providing an interface on the phone.
Shortly after the release of Loop, Android APS implemented the OpenAPS algorithm on Android smartphones. Meanwhile, someone else wanted to implement automatic boluses with the Loop algorithm, so they created a “fork” of Loop called “FreeAPS.” FreeAPS allowed you to use configurable auto boluses to bring glucose to target. Eventually, Loop integrate this feature into the autobolus branch.
Then, the person who had created FreeAPS wanted to implement the OpenAPS algorithm in iOS, so he created “FreeAPS X.” After development stopped, another programmer created a fork of FreeAPS X, called iAPS. The diabetes community liked iAPS but didn’t agree with the software engineering practices, so they forked iAPS and developed another app called Trio, hosted on the Nightscout github.
The OpenAPS algorithm now implemented as OpenAPS/FreeAPS X/iAPS/Trio has worked to support unannounced meals. Loop is very deterministic, meaning what you enter is what you get, and it adapts to a certain extent but is not designed for unannounced meals. I’ve been going back and forth between iAPS and Trio, giving people the choice or offering both. Some Loop developers are on Trio and are working to keep drivers consistent in the two systems.
Marissa: What devices do you need for the different systems?
Rayhan: In the United States the open-source algorithms support old Medtronic models, Eros and DASH pods. There is an iAPS branch with support for DANA pump. AndroidAPS, used more globally, supports a lot of different pumps – DANA, Accuchek, Eros/DASH pods, Medtronic pumps, Diacon G8, Eopatch, Medtrum, and Equil. What people have access to depends on where they live in the world and what is paid for by their insurance if they have insurance.
Marissa: What is required to set up these systems?
Rayhan: It requires some software setup, but once you learn how to build it, it’s relatively easy to replicate. The biggest limiting factor is having enough healthcare professionals to tune the system for optimal use. One of the things I need to work on is disseminating knowledge on how to setup and tune these systems. You also need a smartphone with a battery that will get you through the day and maintain connectivity to pump and CGM. Luckily, you can get an iPhone SE on Amazon these days for $100-200.
Marissa: How many people have you trained on systems like these?
Rayhan: At this point, I have done over 4,000 starts for people with diabetes, ages 3 to 75. I’ve trained folks from higher socioeconomic levels and education to migrant workers struggling to achieve glycemic goals. People who have less experience with diabetes technologies are easier to train; what’s challenging is un-training certain behaviors like micromanaging. The other challenge is weight gain, which can occur once glucose control improves.
How it’s been going
Using the iAPS system has been great for me. My time in range (TIR) has increased significantly and, as I have lost weight, my insulin needs have decreased; it has been easy to adjust the system. For the first couple of days, after I inject tirzepatide, I use a reduction of 75%, then gradually increase back up to 100% at the end of the week. This has been super simple to do and incredibly helpful in preventing lows.
The feature of Nightscout’s Autotune is also embedded in the app, providing updated recommendations for settings that you can review in settings and adjust as needed. The combination of tirzepatide and iAPS has helped my diabetes management in a myriad of ways, and I plan to continue using them for the foreseeable future.
In Conclusion
CWD believes in choice for people with diabetes, and Dr. Lal has worked diligently to ensure people have access to the technologies they would like to use. Given the current global landscape for diabetes technologies, it seems that for many, there is a place and need for open-source AID systems.
Written and clinically reviewed by Marissa Town, RN, BSN, CDCES
