As many individuals may have seen, Dexcom have been increasing their product line primarily based on the G7 {hardware}. The substitute for his or her extra finances gadget, the One, is the One+, they usually’re utilizing the identical {hardware} for the Stelo, the primary US over-the-counter CGM.
While the One+ isn’t but obtainable within the UK, I used to be fortunate sufficient to be offered one by buddies in Europe, in addition to a G7 to run it alongside. Thanks Kamil and Inge.
As you may see from the above picture, they’re mainly the identical, and I used to be ready to make use of two variants of xDrip to seize the info.
Each had been utilized side-by-side on my arm and the intention was to analyse the distinction between the datapoints of the 2 units, quite than in contrast each to fingerpricks.
Suggestions from people who have scanned the firmware is that that is additionally nearly equivalent
Sadly, I solely acquired round 6.5 days from the G7, as a result of one of many edges coming unfastened and sticking to the sleeve of a jumper, which resulted in an unintentional elimination.
This resulted in 1,713 datapoints.
The evaluation
With each sensors utilized inside a minute of each other, the clocks had been nearly synchronised and produced datapoints on the 5 minutes marks of the hour (:05, :10, and so on) making comparability comparatively straightforward.
I exported the datapoints for every sensor from xDrip and aligned them utilizing Excel. I then calculated the relative distinction and used the statistical features in Excel to analyse the info.
Relative distinction is:
100 * [ (yCGM1 – yCGM2)/mean(yCGM1,yCGM2) ]
The place yCGM1 and yCGM2 are the 2 datapoints from the CGMs on the identical time.
Outcomes
If we plot the info on a graph, we are able to see that there’s not a lot distinction between the 2 sensors for almost all of the info set. It’s solely actually on day one which there’s something noticeable.
Zooming in on day one, the G7 seems to have had a barely completely different expertise of insertion in comparison with that of the one. I wasn’t evaluating information with fingerpricks, so there’s not a solution to the query “Is yet another correct?”, nevertheless it highlights that day 1 information may be unreliable and range between sensors.
Trying on the relative distinction , the descriptive statistics advised that we didn’t have a standard distribution, which is backed up by the beneath histogram.
This time, there seems to be a bimodal distribution, with, as soon as once more, the day one variation proven influencing the detrimental tail.
The imply worth of relative distinction was -0.169% and the usual deviation 15.965.
Plotting the relative distinction of the interval of sensor put on demonstrates the place a lot of the variation occurred. Including bands for ±15% additionally demonstrates that the overwhelming majority of the distinction was nicely inside one normal deviation.
There’s additionally a big distinction on day two, which corresponds to roughly 1.30am and is assumed to be a compression low on the G7.
Dialogue
The important thing takeaways from this evaluation are that, as anticipated, the 2 sensors are similar to each other in outcomes, which was anticipated.
Basically, values had been principally inside a spread of ±15% of one another, which is nicely throughout the working vary that Dexcom would anticipate from a single sensor.
A lot of the variation (and due to this fact skew) within the statistical evaluation is pushed by the numerous variations on day one, which highlights the sensor to sensor variation that folks see on the whole after insertion.
In any other case, the important thing takeaway right here is that, as you’d anticipate, there’s subsequent to no distinction between the G7 and the One+. Which is strictly what we need to see.