When choosing an artificial pancreas system, you have a number of choices to make.
DIY or commercial?
Hybrid, or fully automated (if/when available)?
Open, or closed looping?
As discussed in the previous chapter, there’s a variety of different types of APS and related systems, which I won’t rehash here. I also won’t go into details on all of the specific commercial systems. Instead, here are some things to think about to guide your choices.
The most important thing to consider is what matters most to you. You may have strong preferences about the manufacturer, type of pump, type of CGM, or even the size, shape, and colors of systems. Other choices can include some of the following.
CGM Some systems will have only one CGM choice: choose the system, and it comes with the de facto CGM. Others will have options for CGM. Eventually, and ideally, most systems will come with a choice of CGM.
Pump body You may strongly prefer the look of one pump over the other. You may want a tubeless option, a small pump, or a pump with a large reservoir. You may want a waterproof pump. You may want an option that is less likely to alarm going through metal detectors. You may care about how the pump is charged, or what batteries it uses. You may care about which sites work with the pump.
Controller The controller may be integrated into the pump body. Or, it may be a separate device - this could be a standalone, locked-down controller that cannot be used for anything else. Or, it could be something that can be installed on an Android or iOS phone. You may feel strongly about having to carry “one more thing”. Or, you may not trust the security or reliability of your existing mobile device (or the Bluetooth connectivity) to act as the controller for your APS. You may prefer to power and charge the controller separately, or you may find you are better able to keep it running if it’s on your primary phone.
Most of the first-general commercial systems will focus on the above components. But the algorithm and its capabilities, plus the advanced features and other flexibility of usability choices may (and in my opinion, also should) influence your choice of system.
You may feel strongly about using a particular algorithm, which you may have used before from a clinical trial, read studies about, or used in a different system. Some algorithms may have different features than others.
Note: don’t focus too much on things like “basal-only” systems vs. “microbolus” systems - those are just different technical terms for how insulin gets delivered, and are often used to distinguish versions within the same family of algorithm. You should also evaluate the algorithm’s ability to reach blood glucose targets, time in range targets, etc., which will be covered later.
If you choose a particular commercial system, and they have an improved or updated algorithm, how will you get updates or the next version? Will it cost more? What happens if this is inside or outside of your device warranty cycle?
Both default, regular targets and temporary targets may matter to you. The regular or default target can make a big difference. The first commercially approved AP system targeted 120 mg/dL, and the only alternative was a temporary activity target of 150 mg/dL. That may be ok with you, or you may want different targets and flexibility.
In the DIY community, all systems have flexible targets. We use temporary targets to adjust insulin delivery leading into and during/after physical activity. We also use lower than usual targets as a way to adjust insulin delivery prior to a mealtime. Flexibility in targets is very useful for eliminating many every-day spikes and drops, even with a second-generation-capable APS.
If you dislike having to handle your pump or have situations for work or otherwise where you might want to be discreet, consider whether you have to enter carbs and adjust targets on the pump. Or can you use the secondary mobile device, or your own mobile device? This could mean your mobile phone (Android or iOS), or a smartwatch.
Remote monitoring can be useful for many reasons. (As an adult with diabetes, I still find it useful to have my loved ones be able to remotely monitor as needed.) Will your system of choice have remote monitoring capabilities built-in? If not, does it work with one of the DIY remote monitoring capabilities?
Remote monitoring can also enable your own “glanceable” displays on your phone or smartwatch of choice. Think about widgets for your phone or your computer, your watch, and other devices such as your car. Will you be able to access your data in real-time and get it to the device you want?
And, what data will it display? Is it just CGM data? What about knowing if your closed loop is currently automated and working as expected? What about predictions and alerts? What about your net IOB?
You may also want to find out about the management of your APS as it relates to your CGM. For example, does the system stop you from looping (and fall back to manual pumping) if you don’t calibrate your CGM per the recommended schedule? How does it warn or alert you to the need to calibrate, or the end of your CGM sensor?
How does the system deal with the following real-world situations?
● If you forget to bolus for a meal (but usually do)
● If you forget to enter your carbs for a meal (but usually do)
● If you ate, but threw up (food poisoning, etc.)
● If you have gastroparesis and have delayed digestion
● If your pump site has fallen out, and you haven’t gotten the insulin delivered that it thinks you have
● If it’s been learning that you’re resistant because your pump site is too old, how do you tell it when you’ve changed your pump site? Does the learning algorithm take that into account, or can you tell it so?
● What happens when it breaks? How will you get support, or a replacement? How quickly can you get a replacement? What does manual mode look like after you’ve been on your APS?
● Will it alert you if you need carbs?
● Will it alert you if it can’t give you more insulin, or if it’s been giving you the maximum amount it can give you for a certain period of time?
● Is this device likely to add to your diabetes burden or reduce it? Ease of use is important.
One helpful resource for answering some of the above questions about coming-soon commercial devices is Diatribe’s list of upcoming US and EU-based artificial pancreas systems, which you can find at https://diatribe.org/artificialpancreas. Not all manufacturers will have released enough data and information about their systems to answer the questions above, but hopefully when a system is released to the commercial market, there will be enough information about the above to help you make an informed choice. If the manufacturers don’t release enough information to answer all of the detailed questions, you might need to wait until someone in the community gets their hands on one and writes up a more detailed real-world review.
I’ll use some of the examples from the DIY world to show you how answering of the above questions may drive your APS choices.
For starters, you need to know your choices. In the DIY world, that’s generally OpenAPS, Loop, and AndroidAPS. At a high level:
OpenAPS works with older Medtronic pump models, can be used with any CGM, uses a separate micro-controller “rig”, and can be used with either Android phones or iPhones.
Loop has a different algorithm than OpenAPS, also works with older Medtronic pump models, can be used with most CGMs, uses a separate radio device (called a RileyLink) to bridge communication to an iOS device, and uses an iOS device (usually an iPhone) as its controller.
AndroidAPS uses the OpenAPS algorithm, works with DANA*R(S) or Roche Combo or Insight pumps, works with most CGMs, and does not require a separate radio device or controller because the controller is an Android phone.
(Omnipod pump driver will be coming soon to Loop and also probably AndroidAPS, but is not released in the master branches of either system at this point in time)
Generally, all the DIY systems are similar hybrid closed loop systems. They’re designed to send temporary basal rate commands to the pump. They have hardware and software safety limits for insulin dosing, which you can adjust based on your preferences, and change those over time. You can set and adjust your regular targets in addition to setting temporary targets. If they break, or the controller battery dies, or it fails for any number of reasons, at the end of the temporary basal rate, they fall back to standard pump operation.
You have access to an older Medtronic pump, but don’t care which phone you use. You can choose between OpenAPS or Loop (and in the future, AndroidAPS, too), depending on which algorithm you want to use and whether you want your controller to be on an iOS device or not.
You only have an iOS device and an older Medtronic pump. Your choices are still between OpenAPS and Loop - both can use an iPhone. It’s then a question of whether you like Loop’s algorithm and interface (and you’ll carry a RileyLink), or OpenAPS’s algorithm and using a similar sized rig (to the RileyLink) to be both controller and radio.
You have a Roche Combo or DANA*R pump, and an iPhone. Until a device driver is written for those pumps and the iPhone, you can either switch primary phones to be an Android device, or choose to buy a small, secondary Android phone to be your controller, and in either case use AndroidAPS.
You like the OpenAPS algorithm, and have an Android phone. In this case, you can choose between AndroidAPS (if you have or can get a compatible pump), or using OpenAPS (if you have or can get an older Medtronic pump). OpenAPS is rig-based and doesn’t require a particular phone type.
Here are some other things you should consider when making choices about your APS:
1. Picking one kind of technology does NOT lock you into it forever. If you’re DIYing now, you can choose a commercial system later. If you start on a commercial system, you can still try a DIY system. Also, you can add on components of DIY systems (such as remote monitoring, or Autotune reports) with a commercial APS.
2. Don’t compare the original iPhone with an iPhone X - compare apples to apples. Let’s be blunt: the Dexcom 7plus CGM is a different beast than the Dexcom G4/G5. And the G6 is different from those. Similarly, Medtronic’s original “harpoon” sensor is different than their newest sensor tech. The Abbott Navigator is different than their Libre. Don’t be held up by perceptions of old tech – whether it’s CGM, pump bodies, or algorithms. Make sure to check out the new stuff with a somewhat open mind, and truly compare apples to apples. These things in commercial systems should change over time in terms of algorithm capabilities, targets, features, and usability. They certainly have in DIY – we’ve gotten smaller pancreases, algorithm improvements, all kinds of interoperability integration, etc. If you’re reading a review online about someone’s experience with an APS, make sure to pay attention to which version they’re using. A lot may have changed since the review was first written.
3. All systems (DIY and commercial) have pros and cons. They also each will have their own learning curves. Some of that learning is generalized, and will translate between systems. But like switching between iPhone to Android or vice versa – your cheese gets moved and there will be learning to do if you switch systems.
4. Remember, everyone has different preferences and choices – and everyone’s diabetes is different. Figure out what works well for you, and rock it! And when your life changes or your preferences change for whatever reason, know that you’ll have choices then, too. If for now, you don’t think you (or your teen, for example) would remember a separate controller or be willing to carry it around with you, an all-in-one system where the controller is embedded within the physical pump might be ideal. If later you decide that remembering to carry something with (or something else on) you is worth the benefits of another system, you may decide to switch systems.
5. Focus on what outcomes are most important - and apply - to you. For you, the ideal outcome may be less work on your diabetes. It could be a different A1c, or an improved Time in Range (TIR) goal. However, while some of the research studies will talk about A1c and TIR and many people will share theirs, remember to also ask about what targets the person has set, how much time they spend closed looping, and how much work it is to use or maintain their system. One metric may be most important to you, but be careful to not use one metric that is not representative of the entire APS experience when making (or sharing) your choices.