G6 calibrations not picked up

[rumil]

Even though G6 does not require calibrations, sometimes it is still useful to calibrate it.
When entering calibration in Dexcom G6 app, Loop doesn't seem to intercept it correctly.

I did a calibration in G6 app at 18:13 ("Ostatnia kalibracja" is "Latest calibration"):

You can clearly see in BG graph when this happened:

However viewing G6 information in Loop does not show any calibration events:

Attach an Issue Report
Loop Report 2019-03-05 19_41_09+01_00.txt

To Reproduce
Steps to reproduce the behavior:

1. Calibrate in official Dexcom G6 app
2. Go to Loop -> Settings -> Dexcom G6 - no calibration information is present

Expected behavior
Loop should intercept calibration event

Phone

• Hardware: iPhone X
• OS Version: iOS 12.1.4

Loop Version

• Version Number: 1.9.4
• Repo: katie-anna

CGM

• Device: Dexcom G6
• Manager app: Dexcom app

Pump

• Manufacturer: Medtronic
• Model: 722
• Firmware version: 2.2A 1.1 0B 0B

[shaunlichti]

For what it's worth: I've experienced this as well on the omnipod-testing branch. Can also provide issue report if helpful, though I'm guessing this is a well known issue at this point.

@ps2, I know that momentum effect isn't calculated if one of the last 3 points is a calibration point, but do calibration points impact calculation of retrospective correction at all? E.g. if calibration has caused a 50 point drop in BG during the 30 min look-back window, will retrospective correction consider that a forecast error?

[sraka1]

@ps2 @mpangburn I still see this in dev and autobolus. I then have RC discrepancies of up to 2 mmol. I think this could have a pretty severe adverse effect if the calibration induced RC observed BG is then higher than predicted (delivering too much insulin - hypoglycemia), perhaps Tidepool could address this and backport into DIY. I think resetting RC on a calibration would be a safe bet. (Ignore the Integral part in screenshot...just testing, behaviour is same on all branches)

[elnjensen]

For what it’s worth, I regularly have big calibration jumps like that on G6 as well - factory calibration on a new sensor is almost always too high, so I calibrate on the first few days with big downward drops.

[loyalchallenger]

For what it’s worth, I regularly have big calibration jumps like that on G6 as well - factory calibration on a new sensor is almost always too high, so I calibrate on the first few days with big downward drops.

Same for me, which also results in false dexcom alarms going off all night.

[ps2]

@sraka1 Yes, you are correct that RC is not disabled when a calibration is part of the recent history, and it is indeed something we're looking at changing.

[ps2]

In response to the original question of this ticket, if anyone captures this specific event happening (Loop not picking up the calibration event) on 2.2 or later, with the updated device logs, those logs may point to a fix for the issue. Please attach any log captures to this ticket, post in zulip, or send to me directly on zulip.

[marionbarker]

I captured a log after a calibration event. I’d been out of contact of Loop just prior to calibration. So I captured a second log 15 minutes later (coincident with posted screenshot). Let me know if you want one or both. (These are still running 4+MB each even though I’m no longer running Tidepool code.)

[elnjensen]

OK, attached is an Issue Report from a recent case where I see the original issue - calibration on Dexcom G6 that is not picked by Loop. Calibration was at 5:10 PM EDT (21:10 UTC).

Loop Report 2020-06-09 17:54:12-04:00.txt

[github-actions]

This issue is stale because it has been open for 30 days with no activity.

[github-actions]

This issue was closed because it has been inactive for 14 days since being marked as stale.