Dexcom Stelo Calibration

[aeroevan]

Since the native calibration does not work for the stelo, it would be nice if one of the calibration plugins could be used for this device (the 3 sensors I've used have been fairly bad).

I attempted to enable a calibration plugin, but couldn't get it to apply to the data shown on the app so I might not have set it up correctly.

[PhilTU]

"G6, G7, 1, 1+ and Stelo sensors are factory calibrated, you will enter the calibration code or sensor code at startup. Should you want to calibrate, xDrip+ will only send the calibration data to the sensor transmitter for elaboration: it will not apply any local algorithm. xDrip+ in native mode relies on the transmitter data, this is why your receiver (if you own one) will display the same data."

I'm not a Dex user, this is the only thing I saw that might be useful, use with care :-

Can you turn off "native mode" ? Settings −> G5/G6/Dex1 Debug Settings
−> Native Algorithm

https://navid200.github.io/xDrip/docs/Native-Algorithm.html

[aeroevan]

It does seem like people have had some success with better accuracy with alternative placement locations (I've just tried the back of my arm)

[PhilTU]

I do wonder if they intentionally sell the less accurate sensors as stelo though

i don't know how you could do that, unless you're suggesting a failed batch of standard sensors gets relabelled as Stelo.

These cut down versions are IMHO just a regulatory hack for jurisdictions that require prescription and perhaps the opportunity to provide apps without alarms and the "diabetic axis of failure" for people wanting to closely monitor glucose at physiological levels.

[Navid200]

Manufacturers do binning.
Intel used to screen all the CPUs from a wafer and label them differently based on test results. They would come up with a family of products that were addressing different clock speeds. But, they were all coming from the exact same production line. They were all being tested and binned and sold at different prices.
An extreme case of this was Celeron. There was no intent o create a weak CPU. But, there was some parts with one core failing. They just disabled the failing core and sold it as a very weal CPU with a different name (Celeron). At least, this is what I read. I cannot say this is what Intel has claimed. So, I am not saying I have all the facts about Celeron. But, it kind of makes sense. It was an extremely weak CPU.

I'm not saying Dexcom is doing that. But, I am saying there is nothing illegal about it. In fact, it is done in production lines in order to maximize overall yield.

[PhilTU]

I could do it with a cpu or a resistor. I don't know how I would do it with a sensor easily - dip them individually in a defined glucose solution ? That's why I thought a batch that spot tested out of spec might be regraded rather than on a per sensor basis.

[aeroevan]

That's exactly what they do, and the FDA recently cited them for not accounting for evaporation properly: https://www.fda.gov/inspections-compliance-enforcement-and-criminal-investigations/warning-letters/dexcom-inc-700835-03042025

[i3ullseye]

New poster here, not sure if there is a suggestion forum or something. But I am a Stelo user, and they are frequently 20 off high, or 10 off low. It would be nice if Xdrip offered a manual offset/intercept option. Sure, with all the due warnings, and a toggle to enable it ensuring you are taking all known risks. But then, for the current sensor, allow us to enter an offset amount, and just adjust the in app reading plus or minus whatever offset we choose. Wouldn't be as accurate as proper calibrations and sloping, but could get around annoying blanket issues like this.

This would work for ANY monitor then, regardless of their own calibration routines, If the manual entry allowed time entries also, you could even take multiple tests and generate a slope properly.

[i3ullseye]

Ignoring the error..... the Stelo is "Factory Calibrated" they say. Entering manual tests, flagging them as calibrations, does nothing at all. I am not sure I am communicating myself adequately here or something.

Example..... with close to real numbers as I recall...

---I had one sensor telling me I was at about 125.
---I did a finger prick and saw 89
---Using a second Glucometer, I did another finger prick and it said 96

The Glucometers have calibration fluid I use, and both were pretty close in range. I would think that gives me a good view as to my actual Glucose level, roughly 93 or so. This means the CGM is reading almost 30 above that.

I did the same thing a few hours later, with another result of the CGM being 25 to 30 high.

So why can't we, on the xDrip side, shift the displayed result and show a calculated result with a -25 applied? I know that interstitial is not blood, I know that sensor placement can impact things, and I know some sensors can just be faulty. NONE of this would matter if we had the ability to automatically adjust reported result based on calibration inputs. We could even use those faulty sensors, or make up for a bad placement, by using the software to do an adjustment for us.

Yeah, we can do this in our head, but then you have to disable or change alarms, because the system is basing it all on the raw data from the CGM sensor. It wouldn't be perfect, and certainly shouldn't be used if controlling a pump, but for T2D it would make the whole process more accurate to our expectations.

[Navid200]

@i3ullseye This thread was opened by someone else.
I am working with that individual hoping to figure out if a Stelo accepts calibrations or not. Regardless of what you want, having an answer to that question is very valuable.

I welcome discussions and logical debates. Please open a thread and we can talk. I prefer not to derail this discussion.

[aeroevan]

I'm currently out of my stelo cgms, so if anyone else can test that would be nice. Might be a week or 2 until I can test again. Unless those calibration logs would still be available somewhere in the app

[aeroevan]

I was using my contour next one, connected to xdrip via bluetooth but I wasn't using the control solution

[Navid200]

There are no relevant logs.

xDrip has a lot of log content that is outdated. I am trying to clean it up. This is an example:
#3646

The only way to test this is to calibrate following these instructions:
https://navid200.github.io/xDrip/docs/Calibrate-G6.html
That page was created for G6. But, the requirement for small corrections to avoid rejection by the transmitter is valid for G7 as well.
In short, if your reading is 10 and your glucometer measures 6, you cannot calibrate to6 or else, it will not be accepted. You can calibrate to 8 in that case. You will then need to calibrate again later for another small step.

If you submit a calibration and in about 3 readings, there is no change, it means Stelo does not accept calibrations.
But, if you see that Stelo has reacted to your calibration and provides lower or higher readings depending on your calibration, it is accepting it.

Please note that just because you submit a value does not mean that the result will exactly match what you entered.

[thoraxe]

I have just started using Stelo sensors with xDrip and would be happy to provide any information for debugging and calibration purposes. I am not diabetic, so absolute accuracy is not very important, but I am a nerd and want the data, so I'd prefer to be able to calibrate. I have no problem doing finger pricks - I have a glucometer, but I think the strips are expired.
Please let me know how I can assist the community.

[Navid200]

Wait for Stelo to run for 12 hours to reduce the jumps.
Then, when the trend is flat, perform a test with your glucometer just to get an idea if you are lower or higher.

Let's say Stelo shows 110. Let's say your glucometer shows 130. Never mind that it is so much higher because it could be due to the strips having expired.
Perform a calibration by tapping on Add calibration from the main menu (top left on the main screen). Enter 115.

On the other hand if your glucometer had shown a lower value, enter 105.

Wait for 3 consecutive readings after you submit the calibration.
Then, please see if you see a jump in the subsequent readings to the calibration.

[thoraxe]

Will do. The strips expired in January of this year but I'm happy to get more. They're super cheap. It hasn't been 12h but I just did a test and the Stelo is showing a little higher than the strip. But I'll wait to get fresh strips to do any calibration work.