In the first part of this two parter we looked at what technology choices are available or coming to MDI users. In this part, we’ll look at how these tools enable decision support in relation to MDI and what that could mean for users in the future.
At the moment, the majority of commercial tools allow you to capture various pieces of information, entered manually through one or more apps, and see that information on either an app on your phone (LibreLink or Dexcom apps) and/or upload that data to a service like Tidepool, which gives you the ability to review all the data in one place. If you use it with Diabetes apps on an iPhone that work with Healthkit, the mobile app can also upload diabetes data and give you more information.
In theory, you could then look at the data and see where carbs have been entered, boluses, basal rates, and in theory at least, taking all the data together, see what the outcomes were. As long as you stick with an iPhone, and manually enter a lot of information into various apps, or even just one app, it should be presented in Tidepool for easy review.
Similarly, it’s possible to do the same thing with Diasend and also with NightScout:
While a level of automation has been achieved for many of these systems using various CGM and pump data sources, and DIYAPS systems, there remains a large gap when it comes to MDI.
If you want to get the data into these review tools, and subsequently use it, as an MDI user you’ve been required to manually enter carbs and doses of insulin into whichever logging app you prefer, and if we’re honest, logging stuff in an app is little different to writing stuff in a blood glucose diary. Very few of us do it consistently or effectively.
This is where the connected devices start to come into play.
What do the commercial apps offer?
Initially, based on the information that Novo has announced, their product will send data to tools like MySugr, Glooko and Dexcom (and other partners that may not yet have been announced). Likewise, Companion’s inPen is compatible with Tidepool, while the available smart adapters don’t seem to integrate with other systems just yet.
What this tells us is that now, you’ll be able to capture the insulin doses that have been given, and dependent on which system you use, capture that for both basal and rapid-acting insulins. So we’ve made a step in the right direction, but we still have one small problem. There’s no incentive to capture carb information in most cases.
This is where what’s coming from Bigfoot differs.
Bigfoot Inject will offer the following:
And in their own words:
Bigfoot’s vision is to deliver injection systems as a monthly subscription and bring together insulin dosing support and glucose management with the goal of reducing the daily burden of diabetes management for people with insulin-requiring diabetes. Bigfoot also intends to provide straightforward methods for healthcare providers to review comprehensive data by virtue of a connected service, with the goal of enabling care teams to provide more informed dosing regimens. Bigfoot’s solutions have the potential to improve health outcomes and quality of life for people with diabetes and reduce costs throughout the various healthcare delivery modalities.
This suggests that if you subscribe to the Bigfoot service as an MDI user, you’ll get both the connected tools, and importantly, software that provides you with bolus advice around mealtimes, and I hope, additional longer term pattern determination that suggests changes to the basic factors involved in diabetes care, e.g. Carb Ratio, basal insulin requirements and correction factor.
Whilst the device itself looks like it still requires you to dial up your dose manually, by entering carbs into the bolus calculator, you’ll at least be able to capture all the information that’s needed for reviewing the data, as an AI model might, with a single entry into an app that gives you a positive incentive for doing so, as it will be aware of insulin on board, carbs on board and therefore should be able to advise a better dose.
Now this, of course, is really useful technology that allows much greater ability to provide better decision support for a user. But it’s not yet available.
So what is?
There’s next to nothing commercially available that offers this type of interactivity and automation.
There are apps like PredictBGL, which for a subscription will provide you with a level of predictive information based on real time CGM data or blood glucose data and a predictive algorithm that looks at your data over a period of time and then highlights when it thinks you will have a low. It also incorporates a bolus calculator that takes fat, protein and carbs into consideration, and will adjust the dose accordingly. For MDI users on fingersticks, it will also alarm to remind you for a post prandial blood test, at which point additional insulin may be recommended.
Unfortunately, this is all manually entered, aside from the Sensor glucose reading, so you still have an element of logging. If PredictBGL can integrate with some of the upcoming smart devices, then they’d have an extremely useful piece of software on their hands. I’ll put my hand up and admit that I tried predictBGL a long time ago. My biggest issue with it was the manual entry of everything, and I stopped using it before I could see any benefits of the predictive technology.
So is there anything from #WeAreNotWaiting that could help?
Funny you should ask that question.
As it stands right now, Spike, NightScout, xDrip and AndroidAPS all offer bolus IOB aware calculators that can be used to provide a bolus calculation and/or carb requirement. They all use standard diabetes maths with variations on the underlying algorithm relating to insulin on board, with all of them stemming from the original oref0 model from OpenAPS.
While these allow you to create the bolus data in whichever system you are using, only one of these will also integrate with available pens, and that’s xDrip. It provides experimental connectivity to InPen and Pendiq 2.0.
Perhaps the most exciting thing with xDrip is the Pendiq 2.0 integration.
Why does this matter? You’ll note that there’s a “Send treatments to pen” option showing in there. Pendiq 2.0 is a powered pen with bluetooth connectivity, as we mentioned in part 1. So if we enter carbs into xDrip, we can use the bolus wizard to suggest a dose. We can than enter that dose into xDrip and have it delivered to the device for delivery.
At this stage, that still requires two entries in xDrip, but it does mean that the bolus calculator could be developed further to suggest the dose and push it to to the device so that you are now only having to do a single bolus calculator activity to log carbs, get a suggested bolus and have it sent to your device for injection. This has significantly more utility than what we currently see with MDI and provides something similar to the bolus calculator that’s available on pumps.
But that’s only half the story.
The other piece of this is the longer term “macro” adjustment, to try and get a better set of basic ratios in order to fine tune your setup. Again, it’s early days, but Autotune does provide a mechanism for doing so, via a web interface if you’re using some form of CGM. You need a NightScout website, but it’s possible to configure the basic profile in NightScout such that Autotune can interpret it and provide suggestions on how to adjust your basal insulin, carb ratio and correction ratio. Whilst I’m not going to go into the details of how to use it with MDI here, if you’re interested in looking into that, leave a comment suggesting so and we can look at writing it.
Is it all being brought together?
As we can see here, it’s a case of “not quite there yet” for MDI users. There are clear areas where both commercial and DIY groups have put in place steps to try and integrate bolus calculation, semi-artificial pancreas technologies and some level of macro-data determination, but they are all still partial solutions. This is primarily driven by the age of this technology, as much of it is still in its early stages, unlike pumps that have been around for a long time.
With greater access to smarter devices, we’ll see greater integration with decision support systems, and in the longer term, I wouldn’t be at all surprised to see the likes of Tidepool introduce something similar to Autotune, to make recommendations as to how to adjust basic settings, or integrate predictive analytics into the smart phone apps to warn when low glucose might occur.
So while it’s a good start on the MDI front, you’ll need to keep an eye on the various groups over the next couple of years to really start to see the innovation happening.
The devices are just the start…