It’s now just over a year since I first built myself an OpenAPS rig and entered into the world of the Hybrid Closed Loop. In that time, the open source technology has moved forward dramatically, delivering a lot of innovation and most importantly, the ability to step away from managing diabetes. Given this milestone, I thought it appropriate to do a review of that period.
My first rig was built on a Raspberry Pi, using a Carelink stick and the 640G as a glucose monitor to provide the all important glucose data. It was functional, but not really portable. it also had a terrible range, so the purchase of something using a Ti chip became paramount, and I eventually ran a couple of Pis, one at home and one in the office, to provide the functionality. Whilst great, this wasn’t very portable.
I also updated my glucose data source to Dexcom to reduce what I had to carry. And the issues with battery life and portability led me to Loop. The idea of running it all from an iPhone was very appealing, and for a while I used Loop.
One thing I struggled with using Loop was the carb absorption model, and having used AMA in OpenAPS, I found the static carb absorption to be frustrating, and therefore used Loop when I needed portability and offline access, and OpenAPS when in the office and at home. This approach worked reasonably well for me throughout October and November.
I learned, using these two in parallel, that it took me less time to manage T1D with the backing of hybrid closed loops. I no longer had to watch my back in the same way when I ate pretty much any sort of food. Yes, mechanical issues still arose (set failure, for example), but overall, there was less for me to do.
As we went into the end of the year, and Hamshield made the Explorer Board available for use with the Edison, I jumped at the chance of a much more portable version of OpenAPS. And went portable and properly offline with the Medtronic CGM system and OpenAPS on the Explorer board combination.
This was a revelation in the use of OpenAPS (and I’m still using the same tic-tac box!), however, CGM using the older Medtronic system, while convenient, isn’t terribly accurate and his a set of idiosyncrasies that make using it a not so pleasant experience. As a result, I reverted to Dexcom G5 and xDrip+ on an Android phone, and use that as my offline approach.
This is one area where Loop is still way ahead of OpenAPS.
Early in 2017, I was asked to test the Super-Micro Bolus (SMB) functionality of what became known as oref1. It made handling food significantly easier, and when eating low carb, removed the need to ever enter carbs or carb equivalents, and allowed me to run without manually bolusing like this. It also made eating easier again.
Add to that the pushover alerts that alert you that you might need more insulin or carbs because your glucose trends don’t match what the algorithm is expecting, and you start to have a system where you interact mostly be exception.
It was a change in the system that made a huge difference. I’d go so far as to describe it as a significant milestone for me.
In March, I finally got my hands on Fiasp, and it confirmed the suspicions I’d had in October 2016 that these types of insulin would be ideal for use with closed loops. The results were really impressive and I went out of my way to test it out. I wasn’t disappointed.
In more or less this form, I took my pancreas with me to Glastonbury, and it did a stirling job of keeping going, even if the challenge of keeping it charged was a slight frustration. Just something that comes with the territory.
Over the summer, the combined forces of the OpenAPS and Loop development gurus recognised the need for exponential curves for use with Fiasp. After much discussion and a particularly helpful interjection to create some scalable curves, these were coded up in a development branch and I was happy to jump in and test them. For me they made a notable difference to the way that OpenAPS (and as I understand it Loop) handled the way the tail differs from regular insulin. It gives a much more stable result, and will make it into the dev branch of OpenAPS before too long, with exponential curves for both Fiasp and fast acting.
In the meantime, Loop now has a dynamic carb absorption model, overcoming the thing I found hardest to live without when using that platform, and the feedback on it is excellent.
Then there’s what comes next using the capabilities of Fiasp combined with OpenAPS. When it works for you (and it admittedly it doesn’t for everyone) then you can do some very interesting things with timings of boluses and how those might be automated. Some of the things I mentioned in October 2016 start to look possible.
That’s my year of looping, but what has this all meant for me? As I mentioned back at the end of last year, it’s hugely changed how I handle T1D. From no longer needing to accurately carb count, via not mattering if I forget to bolus, to more or less safely ignoring what happens for great swathes of time, these are some of the biggest changes I’ve experienced in my time as a Type 1 Diabetic. And maintaining 85%-90% time in range, and an Hba1C lower than 6% whilst life goes on around me and I focus on it. This is what technology is capable of.
At the moment it’s not the easiest thing to build and get into, but within six months, I expect that many more people in Europe will be able to run a form of closed loop using nothing more than CGM, an Android phone and a Roche Accuchek Spirit Combo pump. That’s still a cost of entry, but not as high or as difficult to obtain.
Whilst it’s not a cure, closed looping is the next best thing we have right now (and whatever anyone says to you, don’t underestimate how important that is). While it still requires some hygiene (calibrating sensors and changing pump sites) and a level of commitment, the #wearenotwaiting community has come together and is making it possible for many more people to start down the line of improving their diabetes care. It is an exciting time right now, and a space that’s well worth watching…