I’ve been using the Freestyle Libre for over a year now and the third party apps are getting better and better. Glimp has now reached the point where it can post data to Nightscout. And very successful it is too. I’ve got my hands on an Alpha version of the code, and deployed it, and it all seems to work very well.
Now, parents can get teachers to scan using an android phone and see what their kid’s glucose levels are throughout the day. One might argue that this could be considered overprotective on the part of the parents, surveilling their kids 24/7, but that’s not a discussion for this post.
What’s interesting is that Libre data is now transmissible to NightScout and the interesting factors that this chucks up.
So let’s take a quick look at a Nightscout print:
This will be familiar to users of NightScout. A series of dots showing the data points. It’s perhaps less familiar to those using the Libre, and highlights the major difference in operation between the Libre and CGM. If you haven’t scanned for fifteen minutes, the data on Libre is 15 mins apart. On a Dexcom, it’s every 5 mins. Regardless.
If we take a look at my overnight trace, we see many more single dots and then a cluster:
What does this mean?
Well, if we review how the Libre works, it collects data every minute and then effectively conflates it to 15 minute interval data. As a result, you only ever see a point in time reading for every fifteen minutes more than 15 minutes after the last sensor scan. Both of these occurrences can be seen in the picture above. The highlighted cluster is the data between the end of the last 15 minute interval and the sensor scan.
Scanning the Libre effectively resets this interval, as the below picture shows:
The period between 7am and 8am has multiple data points taken at a minute basis as a result of scanning more frequently than every 15 minutes.
But what do these pretty graphs really show us? Well, I guess they show us what we’ve chosen to forget about the Libre. The data, when not scanned, is really only captured every fifteen minutes, and the pretty graphs that the Libre app suite, Liapp, Glimp and the scanner all produce are interpolated from the 15 minute point in time data, helping us to imagine what a 100% trace looks like, but which isn’t really in existence.
Sure, it’s good enough and allows you to manage your diabetes much more effectively, but it is worth remembering how the system works.
And of course, you get all the reports that NightScout provides, which look remarkably similar to those in the Libre package!
Ultimately, it’s a step in the right direction, and it’s what I was asking for in the midst of Diabetes Awareness Month. It is something that could be easily usable by a clinic to keep an eye on what a patient was doing, if they so wanted. Well done to the team at Glimp for getting this into the platform and working. The next question is whether the LibreLink app will be even half as capable? Guess where my money is going…
Love this, any thoughts about building a NFC reader that can be worn on top of the sensor permanently reading the Libre every 5mins and relay to mobile?
I asked the same question when the Libre first came out & was told that it would reduce the life of the Libre battery or similar. I still think it should be possible – perhaps reading every 15 minutes – no different to manual scanning. It would then open up the possibility of alarms making it of more interest to me. I use a Dexcom G4.
There's some discussion of this on the Freestyle Libre facebook group. The issue is activating the NFC transmission.
If you could link up the XADOW NFC and communicate via Bluetooth to a phone, then in theory you could place a reader over the top of the sensor. I think you'd need to cycle the NFC field on and off on the device to pull the data out of the sensor, and as Bob says, it's likely to reduce the battery life of the sensor array.
Having said that, scanning 30-40 times per day doesn't seem to reduce the life of the sensor, so having an array that pulled every half hour doesn't seem like an unreasonable draw. I think every fifteen mins shouldn't tax the battery too badly.
Having read this paper: http://www.ti.com/lit/an/sloa184/sloa184.pdf it looks as though you'd need to cycle the NFC active component on the reader to induce a transmission from the sensor, which can't be that difficult.
I guess the question is more likely to be the size of the product rather than making it work like this. When you consider the size of a Dexcom or enlight transmitter and sensor, you're going to end up with something substantially larger.
A reading every 15mins would be very usable, and BG readings could be used in a open closed loop system like HAPP https://github.com/timomer/HAPP
They could. There seem to be a couple of ongoing projects on this – someone working on an XADOW solution and someone in Italy coming up with something else. The problem with both is that they are both resulting in a fairly large and unwieldy results. Whilst the sensor is fairly flat, the NFCtoBluetooth bridge and battery is the size of a small cellphone. As others have said, you might as well buy a cheap, small NFC enabled cell phone and use that. At least it has a built in NFC platform and Bluetooth, and enough of a CPU to be the brains for an OpenAPS or HAPP solution. It still has to be held over the sensor though (in a runners armband for example).
While extending the NFC sensor to Bluetooth to a phone might be a cheaper way to create CGM, I'm not sure that it's the right way to do it. If we're going to do something like this, I think it might be preferable to work out how to modify the sensor itself to do transmission via bluetooth and still use the applicator.
Any idea if there is any other way of getting minute by minute data for the libre? Mixture of minute by minute vs 15 interval data makes any form of data analysis fairly complicated. Loving your blog by the way, a lot of great and very useful insight!
Unfortunately the sensor only gives minute by minute data for 15 mins at a time, so you need to scan it every 14 mins 59 seconds to guarantee the data.
Glad you like the blog!
How have I missed your blog before? I've got some reading ahead of me.
About the NFC device, a lady in NZ is currently looking at building one, she seems confident.
In all fairness, until earlier this year there wasn't a lot in it. When November hit I decided to write at least one blog post per day (which I achieved and realised that it was the same number of words as a PhD thesis). An average of a post a day throughout diabetes awareness month.
Since then I write when I want to add stuff or something in the news catches my attention. I've a couple of topics lined up.
Hope you find it interesting!