My Summer Project: Building an Artificial Pancreas

By Katie Aldridge

My name is Katie. I have type 1 diabetes. I’ve always worked hard at managing it, but after 23 years, a person gets weary. It requires constant attention to really control it well. Insulin can be quite volatile, and it’s usually not possible to do just the perfect amount every time.

Instead, I make lots of tiny corrections. Too much insulin = eat more. Too much food = more insulin. It sounds simple, but it requires decisions and calculations all day, every day. And all manner of things can interfere, like stress, hormones, lack of sleep, exercise, being sick, etc.
Hello parenthood!!!

Using a Continuous Glucose Monitor (CGM) and an insulin pump help me make these corrections with more precision. My control is good, but I tend to have too many lows. Low blood sugar can range from annoying to dangerous, so my goal is to have as few lows as possible while keeping my A1C under 6. It takes a lot of work. 

At the beginning of 2017, I started seeing unbelievably flat CGM graphs on Facebook with claims along the lines of “Just ate pizza. Look at my perfect blood sugar. This is so easy.” Umm, what??

This was a siren song I couldn’t resist. I started lurking on a FB group called Looped and began reading the online documentation for something called OpenAPS. It turns out there are people designing, using, and sharing open source Artificial Pancreas Systems, which help with things like keeping your blood sugar more level after eating pizza.

An APS is essentially a computer that does all those micro corrections for you. It bridges the gap between CGM and insulin pump. It uses the data from both devices to assess trends, make predictions, and give more or less insulin depending on whatever you need…. EVERY 5 MINUTES.

After reading for months and gathering all the stuff, I finally jumped in and built a DIY OpenAPS. It consists of an Intel Edison (tiny computer), an explorer board (connects wirelessly using wifi, Bluetooth, or radio), and a lipo battery. Throw in a small 3D printed case, and you’ve spent $180. Cheapest medical device I’ve ever owned!

It took me two weeks to set it up and install the software. My kids would be in bed by 8:00, I’d be done with household chores at 9:00, and then I’d work on this OpenAPS project until midnight. I could work like that for a few nights in a row before I’d be too exhausted to keep going the next day. So I’d take a few days off, then try again. This project had a steep learning curve, and it took more time than I’d expected.

I think I encountered or caused more problems than most people do. I knew NOTHING about Linux or a command line, and my particular setup had some uncommon issues. Fortunately, wonderful people from the OpenAPS community helped me on Facebook or the chat room on Gitter.

But getting help takes patience and humility. In the minutes/hours/days between getting an error and getting help to fix it, there were definitely tears and frustration. Then when things went well, I was dancing around the living room in victory. So many emotions! I was surprised at how emotional this entire project was for me.

I think it comes from this HOPE I allowed myself to have. Believing that something like this exists, that it could lighten the burden of type 1 diabetes, and believing that I could build it… all of that made me vulnerable and emotional. My husband helped keep me steady. When things were going badly he’d say, “It’s ok. Of course you’ll make mistakes. You’re learning something new.” He’s awesome.

I “closed the loop” full time at the end of June, but the workload didn’t really lessen until the end of July. That whole month I spent adjusting my pump settings, debugging a big problem I had, and learning how to monitor and maintain the system.

In the midst of all this adjusting, I began having some success. One night I slept through a pretty serious low blood sugar (despite alarms going off). OpenAPS brought me safely back into my target range while I slept. While I taught my music class, I completely ignored my blood sugar, and after class I was a little higher than my target but not much. The (good) stress and activity of teaching usually causes me highs or lows.

Now OpenAPS has my back. Amazing! To have a little computer buddy covering for me is like… grace. Did I correct a low with too many carbs? OpenAPS can fix it. Did I count that meal incorrectly? OpenAPS can help.

After six weeks of “looping,” I ran some reports. The results: an A1C under 6 with HALF as much time spent in the low blood sugar range. I definitely don’t have a flat CGM line, and I haven’t tried pizza yet. But these results have convinced me that this crazy thing I’ve done is actually working. I’m hoping for even more progress in the months to come.

Artificial pancreas is really a misnomer, because a natural, fully functional pancreas does so much more than this. APS technology is not a cure. It’s not something you can set and forget, at least not yet. It has required a huge time investment, so it’s not for everyone.

Even so, OpenAPS is something I’m so excited about, and it’s already improved my quality of life. I am amazed by the incredible minds behind these DIY diabetes technologies. The amount of work put into the Nightscout community is staggering to me. This is a community of people motivated by need, by love, and by a desire to make the world better. May God bless their efforts! And may the diabetes industry follow their lead.

Thank you for reading about my journey. If you want to learn more here are some links: