Accuracy of CGM (Libre 2 vs KetoMojo n=1)

#1

So I started comparing measured values from my Libre2 and ketoMojo which I only do less than daily. After a call to the manufacturer they suggest taking the libre2 measurement 5 minutes AFTER the blood stick. I did just before and 5 minutes after and found negligible discrepancy since I try to only include measurements when my blood sugar was roughly stable. This graph shows ALL measurements to date

image
image.png665×612 28.8 KB

Each color represents a different CGM device. If we zoom in a bit.
image
image.png679×588 27.9 KB

The dashed red lines represent +/- 10% around the central solid line which is perfect agreement,
Most data fit within the 10% error bars and the grand total average of all measurements are nearly identical.

However, the VARIANCE of the KetoMojo measurements is much greater (details at end for stats nerds) —the response of the libre 2 has a narrower range.

There is no significant bias (high or low) but it does seem the cgm is less sensitive in scale. Of course it provides high frequency information which i find very interesting. And then again, who is to say my keto mojo device isn’t wonky.

FYI

Stats
t-Test: Two-Sample Assuming Unequal Variances

Variable 1	Variable 2

Mean 105.4444444 102.5555556
Variance 384.6539683 102.1396825
Observations 36 36
Hypothesized Mean Difference 0
df 52
t Stat 0.785614764
P(T<=t) one-tail 0.217829182
t Critical one-tail 1.674689154
P(T<=t) two-tail 0.435658365
t Critical two-tail 2.006646805

(Bob M) #2

Since the Libre uses interstitial fluid, there’s supposedly a delay relative to blood sugar level. I always thought it was 15 minutes, not 5. But if you’re low carb or testing when your blood sugar should be “fixed”, it shouldn’t matter.

A good test for this would be to eat high carb and see if there’s a delay between when the pinprick goes up and when the Libre goes up.

#3

Makes sense since CGMs aren’t actually testing blood.

#4

Right. Just that I don’t like to pinprick that often!