Libre questions.

Librelink is the 'official' Abbott app. Basic but does what you want I think.

kev-w said:

This is so frustrating!

I'm finding my Humalog to take different lengths of time to start working, but as I'm multi tasking I forget to mark the notes, there must be some editing software out there?

Unless I'm being thick?

Click to expand...

I put all my notes into the MySugr app, but have to do it manually.

I use the LibreLink app on my phone. Can do everything that you are wanting to do @kev-w

The only way I can see of doing what you want is to export the data using the libre software on the PC, in to an excel spreadsheet and then adding in the missing data, at the end of each day? Week or sensor? it can be abit fiddly! Can't you be a bit more disciplined and make sure you always enter the details as and when it happens?

lol. PC based I am unsure mate. Give your DSN or Abbots a call and ask the question. Pretty sure they will be able to help out

Ah, sorry - I tend to assume most folk use a smartphone or tablet these days. I only use my PC to download torrents and upload my scanner now!

For Windows PC I just connect with the cable to the Libre and use the application you find on the Libre already (its like a USB drive, upon where you both have the web app and the data stored). Its a 'browser based' app, so you use it by opening the webbrowser on your Windows PC.

I'm currently working on a tool that if it works as intended will not only do readings for you and take notes of your injections and what you consume, it'll also return a predicted blood glucose graph for the next few hours based on your data. If I can get it to work, it won't be that functional to begin with but the plan is to eventually be able to not only predict how your blood glucose levels will be in the next few hours based on recent data but also be able to recommend injection dose based on what you are about to eat and things like that.

Disclaimer: Libre-wearing Computer Science student

Kev, how long have you had the Libre now?

I am trying to remember back two years to when I first got my Libre, and if I remember correctly, the first few times I fiddled with that standard Libre software on the PC I was bitterly disappointed.
Then after the first week, extra 'stuff' became available.
I can only assume (remember this is a 2 yr old memory) that it took a certain amount of collected data before it could collate to a spread sheet.
Or something.

Sorry, it is just so long ago.

Ah. Yeah. It was a bit much for me too, and I don't have insulin. I certainly found the whole thing a revelation.

I would try re-connecting it to the computer (it could probably do with charging as well, since mine needs a recharge at around Day 10), and just have a play with the software.

Just keep investigating and running the 'reports' and fiddling around with it.
The spread sheet is pretty basic, but it could be adapted to add your personalised columns, and diary items, and then you can turn that into a graph too.

When I got the Libre I played endlessly (cos I'm a nosey ******), and set up all sorts of things.
But nowadays I just keep to the basics - which for me, are reactions to particular foods in particular portion sizes, and how long it takes to get back down afterwards.

Hope that helps!

kev-w said:

.

Excel, yes I'll have a look at that, "a bit more disciplined" is a fair point.


I have a piece of bundled desktop software, but it doesn't 'seem' to have an edit function.


It was Commodore Basic in my day and your project sounds interesting, although this luddite doubts if mathematical modelling can predict future glucose levels due to the number of variables involved good luck with it though!

Thanks for the advice and suggestions y'all.

Click to expand...

It's definitely impossible as of today to get 100% accurate results due to immeasurable data required(i.e mental state and the like), getting accurate enough results for daily use much like Google Maps is definitely possible though if the data is detailed enough, meaning correct nutritional data as well as mapping how fast certain food types will act and things like that. Fortunately, others have done the actual research on these separate things, I am no mathematician or doctor so my role is strictly to try and put their research together into something useful.

jlarsson said:

It's definitely impossible as of today to get 100% accurate results due to immeasurable data required(i.e mental state and the like), getting accurate enough results for daily use much like Google Maps is definitely possible though if the data is detailed enough, meaning correct nutritional data as well as mapping how fast certain food types will act and things like that. Fortunately, others have done the actual research on these separate things, I am no mathematician or doctor so my role is strictly to try and put their research together into something useful.

Click to expand...

Have you had a look at xDrip+? It's open source, written in Java.

https://github.com/NightscoutFoundation/xDrip?files=1

It's got predictive simulations which I've found to be pretty useful for getting a broad indication of where a particular carb/bolus might take me - there's been a few times I've made a 1 or 2u adjustment to a dose, based, in part, on how the prediction looks, and it redraws the prediction when it gets more data each 5 mins from the cgm transmitter.

Obviously, as you say, there's so many variables involved in a bolus decision, it's never going to be 100%, but I've found that the graph wil give me some useful clues which I can take into account in the decision making process which I might otherwise wouldn't have had, so I'm a big fan of it.

There's customisable parameters for ratios and five or six other things.

The data source can be dexcom, libre/blucon and the many home made transmitters such as limiTTer.

Quick screenshot below to give an idea of what's going on. The blue dots are the cgm trace, the large purple dots extending out after the blue ones are the prediction based on the entered carbs/bolus, the green lines are the insulin levels decreasing over time, the small purple dots are the carb absorption.

Scott-C said:

Have you had a look at xDrip+? It's open source, written in Java.

https://github.com/NightscoutFoundation/xDrip?files=1

It's got predictive simulations which I've found to be pretty useful for getting a broad indication of where a particular carb/bolus might take me - there's been a few times I've made a 1 or 2u adjustment to a dose, based, in part, on how the prediction looks, and it redraws the prediction when it gets more data each 5 mins from the cgm transmitter.

Obviously, as you say, there's so many variables involved in a bolus decision, it's never going to be 100%, but I've found that the graph wil give me some useful clues which I can take into account in the decision making process which I might otherwise wouldn't have had, so I'm a big fan of it.

There's customisable parameters for ratios and five or six other things.

The data source can be dexcom, libre/blucon and the many home made transmitters such as limiTTer.

Quick screenshot below to give an idea of what's going on. The blue dots are the cgm trace, the large purple dots extending out after the blue ones are the prediction based on the entered carbs/bolus, the green lines are the insulin levels decreasing over time, the small purple dots are the carb absorption.

View attachment 25668

Click to expand...

xDrip popped up when I looked at apps to scan sensors but did not know it could make predictions as well, I'll have a look and see how it works, on first look there's a lot of code and files so it's gonna take some time but it definitely looks interesting. How accurate do you find the results?

jlarsson said:

xDrip popped up when I looked at apps to scan sensors but did not know it could make predictions as well, I'll have a look and see how it works, on first look there's a lot of code and files so it's gonna take some time but it definitely looks interesting. How accurate do you find the results?

Click to expand...

As a cgm, it's pretty accurate - being able to calibrate libre output makes a big difference.

As a prediction generator, hmm, accuracy is a relative term! If I enter X u and Y carbs to get an idea of what that looks like before deciding whether to bolus that, it frequently looks sketchy, like it's not got enough information and is still thinking about it, so I don't usually attach much weight to that. But sometimes it'll suggest that a different combination would be taking me well out of range, so it might make me substantially rethink that approach.

The curious thing, though, is that once I've taken a shot/food, then, after about 10 to 20 mins, it'll regularly tighten up and rethink the original prediction. I'm not a programmer, but I wonder whether it's just because now that it's been told I've had Xu and Yg, and also has several actual cgm readings, it just makes a much better stab at it as it has started to test the prediction against reality and adjusted it. I've had a fair few occasions where I've cross-checked against bg in those sort of situations and it has frequentlt made a remarkably good job, which gives me a fair bit of confidence in just keeping an eye in the prediction so I'll get a heads up on what might happen in the next couple of hours, so I've got plenty of time to consider corrections.

It's difficult to explain properly without using it for an extended period. After a while you just kind of "get" what it's indicating and decide how much to trust it. I'd miss it if I didn't have it.