Freestyle Libre delay/extrapolation

[Westley]

The manual says readings may be inaccurate when sugars are rising or falling rapidly (which has been frustratingly often just recently ).
I've also seen several mentions on here of a delay of somewhere between 5 and 15 minutes compared to blood tests.

However, it seems to me that it is not simply a fixed delay. Sometimes I take a reading when it is falling fast and get a low reading of, say 3.0, then a while after I have treated the hypo I scan again and look back at the graph and it then shows the lowest it got before coming back up was 4.0 or more. Same thing with fast rising and high readings.

It looks to me like it is extrapolating. So it knows its interstitial fluid reading is behind the blood, and based on the current rate of change it tries to match what the blood actually is currently. The manual distinguishes between levels measured and levels reported, which would fit this.

Does this match with your experience? (and do you think it is linear or higher order?)

Also, when we look back at the recorded data, are the actual IF readings already appropriately shifted back to give the approximate blood reading at that time?

I'm trying to get a better understanding of the time taken for different doses/foods to affect me, and it's a bit annoying that Abbott aren't open about this, and the meter doesn't distinguish between what it is actually reading and what is an informed guess. I'd rather it showed the extrapolated part of the curve in a different colour at least.

 

[Heathero]

Sorry no idea, didn't know this happens but have just ordered one. Helpful to know this . I was aware that readings may vary from finger tests so will do some comparisons when I get it. Hope someone can advise you, if not believe you can email or phone maybe they could advise. Be good to know if you find out. Re delay you mentioned of 5-15 minute delay is this after blood test or before?

 

[noblehead]

@Westley, when reporting my first sensor the guy from Abbott said you have to leave a 20min gap as the IF reading lags 15-20mins behind a finger bg test.

 

[Westley]

It seems a bit shady of Abbott to say these things off the record in phone support, but gloss over them in the promotional materials and even the manual.
My point is that it does not seem to be simply a case of lag. If that were the case I could take it into account, in combination with knowing what I've recently eaten/injected, to guess where my sugars are now.
The problem is there seems to be a lag which the reader also tries to correct for, often incorrectly, particularly when there is a sudden change in direction, such as when treating a hypo. I'd rather know which part is the real reading so I can take into account eg. the glucotabs I've just eaten in my own extrapolation of the data.

 

[steve_p6]

This is all fairly well understood on the Libre group and also that if you are low/high on Libre with a quickly trending line then it is prudent to do a BG check before taking action. That said once you have been on Libre a while you get much better at balancing food/bolusses/IOB and can try to reduce how often this happens.

 

[tim2000s]

Abbott freely admit it uses the same predictive algo as the navigator but with much better results.

The accuracy study has stats for the lag of roughly 5-10 mins. The G5 Dexcom operates with a similar algo.

This is discussed in the study that you can get to from here http://crick-tech-munch.blogspot.com/2015/11/freestylelibre-freestyle-libre-accuracy.html

I think that they don't mention it because the majority of users genuinely don't care, rather than anything nefarious.

 

[Heathero]

Received it today thanks everyone for the info re lag in reading will make sure I do finger test if high/low. Hope to find it helpful re getting my settings on pod correct as increase in hypos lately.

 

[Thorpe Pigeon ca]

The manual says readings may be inaccurate when sugars are rising or falling rapidly (which has been frustratingly often just recently ).
I've also seen several mentions on here of a delay of somewhere between 5 and 15 minutes compared to blood tests.

However, it seems to me that it is not simply a fixed delay. Sometimes I take a reading when it is falling fast and get a low reading of, say 3.0, then a while after I have treated the hypo I scan again and look back at the graph and it then shows the lowest it got before coming back up was 4.0 or more. Same thing with fast rising and high readings.

It looks to me like it is extrapolating. So it knows its interstitial fluid reading is behind the blood, and based on the current rate of change it tries to match what the blood actually is currently. The manual distinguishes between levels measured and levels reported, which would fit this.

Does this match with your experience? (and do you think it is linear or higher order?)

Also, when we look back at the recorded data, are the actual IF readings already appropriately shifted back to give the approximate blood reading at that time?

I'm trying to get a better understanding of the time taken for different doses/foods to affect me, and it's a bit annoying that Abbott aren't open about this, and the meter doesn't distinguish between what it is actually reading and what is an informed guess. I'd rather it showed the extrapolated part of the curve in a different colour at least.

I am having exactly the same issues. Especially with lows. This is a problem for me because my job depends on my diabetes being well controlled. We're definitely not getting all the information about how the scanner works.

 

[Westley]

I am having exactly the same issues. Especially with lows. This is a problem for me because my job depends on my diabetes being well controlled. We're definitely not getting all the information about how the scanner works.

Hi -
Shortly after I first posted this thread years ago I switched to using Xdrip+ instead of the official Libre app. I find Xdrip much better for getting actual direct info about readings. It does still have extrapolation functions, but in presented in a way that makes it clearer what is actually measured vs predicted.

I somewhat understand why Abbott would want to present things in a simple way to avoid confusion, and just show the reading 'now', without showing anything about how it's actually 'what we predict the blood reading is now based on extrapolating from ~15 minutes ago'.
I think it's a bad decision not to give any other option though as it's rather frustrating when you do want to use it during faster swings.

As I see it, we the users have more information to work with - we know exactly what we've just eaten/injected/exercised right up to the current second, and with experience can make a better prediction of actual current bg levels than the machine can do by simply fitting a curve through a few previous data points of interstitial fluid with its ~15min delay.
The problem is, you have to try and mentally undo the machine's bad extrapolation before you can do your own, and by not distinguishing extrapolation from readings, the reader is actually losing information in what it presents.

I agree it's particularly problematic during lows when trying to judge when it's coming back up. I know people say to use finger pricks before treating during fast changes, but we all know that's not realistic and that most patients do treat highs and lows based on Libre readings all the time. For figuring out the turnaround point during fast changes, individual finger pricks aren't that great either unless you're measuring 3 or 4 times minutes apart.

So as I say, Xdrip was better.
However, that was with Libre1
I've since switched to Libre2, and reluctantly went back to the official app. I really wanted the alarms, and it was easier for sharing the readings with the doctor, and last I checked, it seemed like setting up Xdrip with the Libre2 was more complicated than for 1.
I'm finding it frustrating again though, so would be interested to hear about others using Xdrip/libre2 and what that's like currently.

 

[Westley]

(the other thing I liked about Xdrip was that it would always show readings, even when fast changing or noisy. I hate that the Libre app will 'censor' readings - like 'oh no, it's changing too rapidly for us to give a reading we can be sure is accurate, and it might be an error, so let's just stop telling the user anything at all for the next 20 minutes'. I want to make that decision myself. I understand sometimes there will be noise and errors in the info, and can make an informed decision about how to use it.)

 

[In Response]

@Westley I found xDrip+ easy to set up with OOP2 for Libre2. In fact there is very little to change from the Libre 1 settings as long as you have OOP2 running to hijack the Bluetooth alarm signal.
As before you will not populate LinkView for your doc but you get he calibration, current value on the lock screen, no need for scanning (apart from the start the sensor) and more alarms than LibreLink.

 

[barrym]

Alongside LibreLink I have Juggluco app running, which shows the 1 minute readings sent to LL to give an alarm if required.

It shows a graph of these readings and it is a wildly rough up and down saw-like graph. Very short spikes up or down. When you do a scan, it smooths it out. The purpose I presume is to stop folk over reacting.

If I get a moment later I'll take a screen shot that shows the 'smoothing' and maybe it'll explain what I mean and why you wouldn't really wan to see this.

 

[barrym]

Couldn't wait

Green is scanned curve, Red is 1 minute alarm readings, purple dots are the actual scans.

A couple of instances where you wouldn't have wanted to make an adjustment so LL makes it a bit more realistic.

HTH

 

[ElenaP]

@Westley I found xDrip+ easy to set up with OOP2 for Libre2. In fact there is very little to change from the Libre 1 settings as long as you have OOP2 running to hijack the Bluetooth alarm signal.
As before you will not populate LinkView for your doc but you get he calibration, current value on the lock screen, no need for scanning (apart from the start the sensor) and more alarms than LibreLink.

I have been unable to find OOP2 app neither on iPhone AppStore nor Android PlayStore. Any suggestions, please?

 

[In Response]

I have been unable to find OOP2 app neither on iPhone AppStore nor Android PlayStore. Any suggestions, please?

None of these unofficial apps are approved by the app store or Play store.
If you want to try xDrip+ for Android (there is a different app for iphone), these need to be downloaded from GitHub. The detailed instructions are available from the Files tab of the xDrip Facebook group. This is a closed Facebook group but they seem to accept anyone.
I found this all easy to follow and setup. However, I work in IT.

 

[ElenaP]

None of these unofficial apps are approved by the app store or Play store.
If you want to try xDrip+ for Android (there is a different app for iphone), these need to be downloaded from GitHub. The detailed instructions are available from the Files tab of the xDrip Facebook group. This is a closed Facebook group but they seem to accept anyone.
I found this all easy to follow and setup. However, I work in IT.

Thank you. I will try that.