Whilst Google has recently been in the news a lot in relation to Diabetes, one of the lesser news items that has been published talks about their investment in Oscar, a relatively small health insurance company in the US. But why is this interesting?
“Oscar positions itself as the technology-driven, consumer-focused alternative to the big health insurance companies. It serves as a matchmaker for individuals to get the perfect (and perfectly priced) health insurance policy, and then doubles down by offering fitness-trackers and telemedicine services.” To quote MedCity news.
Why is this important in relation to Google and Diabetes? Google has recently announced a number of partnerships with a variety of Diabetic tech companies, such as Dexcom, that provide inputs into monitoring of diabetes. I’ve not seen anything relating to outputs, e.g. insulin delivery systems, but I wouldn’t be surprised to see that come next.To paraphrase an article in Forbes, Google may not be a device manufacturer but they are good at getting things working together. You can see where this is going. 24×7 monitoring, feedback into the Medical Insurance system and potentially into condition management and all of a sudden, Diabetes can be insured on a “You are what you eat” basis.
I’m not debating whether this is a good or bad model. Personally, I believe that if it is measured on demonstrable behaviours then there isn’t an issue, however outcome based measurement could be more of a problem, especially in T1s. What it does raise is the idea of the interconnected diabetes treatment module.
Imagine, if you will, a world where the bionic pancreas exists. It is an interconnected device that is able to draw on the power of the cloud to manage your body. It secretes Insulin, Amylin and Glucagon and constantly monitors your blood glucose level.
Every minute of every day it runs thousands of scenarios and selects the most likely outcome to deliver either one or the other, or perhaps both. It also measures your heart rate and skin moisture and determines whether you are exercising and what exercise it is that you are doing and adjusts its output accordingly. Based on the historical data it has collected it has generated a model of how your blood glucose changes in response to the stimuli that cause the changes in heart rate, skin moisture and potentially a host of other factors. It constantly updates this model.
Sounds far-fetched? It probably isn’t as far off as it seems. On a regular basis in the financial markets there are hundreds of algorithms undertaking exactly the same kind of process and self-calibration. It’s not such a big deal to make that jump over to human physiology.
But we have to start somewhere. The feedback loop that I’ve described simply isn’t good enough at the moment. The different devices don’t talk to each other. They have no central data store or simple access layer. The delivery devices are simple, expensive and therefore not widely distributed. The standards are limited. A couple of approaches are needed.
On the one hand, creating an “open source pump” would give the mechanism for delivery. A reference design that allowed connectivity via a standard protocol to CGM and fitness trackers combined with a standard hardware design, that, critically, allowed penfill insulin to be used, instead of proprietary cartridges. Something that could be manufactured cheaply, even printed at home, with a flashable ROM that allowed the “brain” to be updated easily.
That can’t be beyond the wit of the Diabetic community, and many in the Nightscout, CGM in the Cloud and xDrip world are actively looking at some of this. In addition, this work, is already creating the centralised, cloud driven database against which the modelling algorithms could be written and tested, then ultimately run.
In the meantime, with the majority of insulin dependent diabetics using MDI and pens, and pen technology starting to include items such as the “remember” functions of the Novopen Echo and third party add ons, it once again wouldn’t be significantly hard to include a bluetooth connectivity layer that allowed this data to be transferred to another device for uploading to the central store every time it was used. Combine this with the ability of CGMs and GMs to upload data to services such as Diasend and by combining the two you’d have the initial Insulin/Blood Glucose measurements being
captured automatically and plotted together. Something that is a very manual process as it stands at the moment, and one that is woefully lacking.
So you have a start. It’s not really a feedback loop, but it is a step in the right direction. The tricky bit is the food. Capturing that in a simple form is much harder. Currently you need to use a manual tool, such as MyFitnessPal (other similar tools are available) and there have been some attempts to use photo recognition as a way of uploading food information. This presents a greater challenge, but I’m sure that with a little ingenuity it can be overcome.
The basic fact is that with the level of interconnectivity between devices that is now available, the official tools to manage diabetes are woefully poor. There are many open source projects out there seeking to address this. A few are listed here:
- Open Source Diabetes – open source Diabetes management software
- Tidepool – an open source data management toolset for diabetes
- The Open APS project – creating an open source artificial pancreas
- Diabetik App – predicts the medication you are likely to use based on your history
- ABC4D – an advanced bolus calculator that is a currently a research project at Imperial College, London
Multiple projects, all looking to achieve similar things but with slightly different goals. If we could get the manufacturers on board, the ease of obtaining the data and therefore the ability to review it and make better decisions would be so much better. This would truly be Dynamic Diabetes Management. Sugar Surfing to the nth degree!
All it takes is one small step…