These are chat archives for nightscout/intend-to-bolus

22nd
Jan 2016
dougtakeuti
@dougtakeuti
Jan 22 2016 01:06
Hi, the loop is crashing because it is out of space. Is this a common issue?
https://gist.github.com/juliatakeuti/7884cba4fe4fb47e209a
Filesystem     1K-blocks    Used Available Use% Mounted on
/dev/root        3651636 3635252         0 100% /
devtmpfs          469748       0    469748   0% /dev
tmpfs             474052       0    474052   0% /dev/shm
tmpfs             474052   12308    461744   3% /run
tmpfs               5120       4      5116   1% /run/lock
tmpfs             474052       0    474052   0% /sys/fs/cgroup
/dev/mmcblk0p1     61384   20312     41072  34% /boot
tmpfs              94812       0     94812   0% /run/user/1000
Ben West
@bewest
Jan 22 2016 01:47
yeah, fairly common
you can try sudo raspi-config
the first option, "Expand Filesystem"
if your disk is actually bigger, it'll expand the file system to match
dougtakeuti
@dougtakeuti
Jan 22 2016 01:57
thanks, now I have a lot more space!
Ben West
@bewest
Jan 22 2016 01:57
apparently jessie comes with bluez 5.27
whichhould work fine s
nice
systemd is definitely starting it without --experimental
Ben West
@bewest
Jan 22 2016 02:13
the ini file that controls that is /lib/systemd/system/bluetooth.service
Ben West
@bewest
Jan 22 2016 02:28
enabled --experimental there
and?
did not find service
Ben West
@bewest
Jan 22 2016 02:39
well it looks like it paired, but same errors as you folks
Ben West
@bewest
Jan 22 2016 03:13
yeah
ok, so jessie has bluetootwth installed by default b
err, not by default
but it does come with 5.27 which does not work
I switched to latest and it does
Ben West
@bewest
Jan 22 2016 03:19
on jessie with systemd, it installs the unit files mostly correctly, except that --experimental needs to be added
Ali Mazaheri
@amazaheri
Jan 22 2016 04:29
ok so what exactly I should follow to start ble on Jessie/Edison :)
Eric
@ecc1
Jan 22 2016 04:29
You also need to do "systemctl daemon-reload" to get it to take effect.
(or just reboot)
Add "--experimental" at the end of the ExecStart line in /lib/systemd/system/bluetooth.service. Then rebooting is the simplest, or you can do "systemctl daemon-reload" followed by "systemctl restart bluetooth.service"
Ben West
@bewest
Jan 22 2016 04:32
hmm
dunno if they compiled it with --enable-experimental
I just dissected a deb package, doesn't look like they enable experimental stuff
pbuilder is happily creating some new rpi environments on my laptop
for edison... are you using that ublinux stuff?
Eric
@ecc1
Jan 22 2016 04:34
Look at debian/rules ... CONFIGURE_FLAGS includes --enable-experimental
Ali Mazaheri
@amazaheri
Jan 22 2016 04:34
yes
Ben West
@bewest
Jan 22 2016 04:35
the one I was looking at did not include that
Ali Mazaheri
@amazaheri
Jan 22 2016 04:35
same setup as @TC2013 and @oskarpearson
Eric
@ecc1
Jan 22 2016 04:35
I did "apt-get source bluez" with my sources from sid, which gave me Debian version 5.36-1
Ben West
@bewest
Jan 22 2016 04:35
wonder what the pbuilder rules for edison are
oh, that's very recent
was experimental enabled?
think that should work just fine
Add "--experimental" at the end of the ExecStart line in /lib/systemd/system/bluetooth.service
this is required no matter what
Aaron Michelson
@Aaybob
Jan 22 2016 05:47
@danamlewis I don't have logs or Nightscout. That's why I want to wear it during the day. I can check on the pump real time what the basal is set to . The "reason" on screen is interesting. Would there be an SSH app for iphone? The NS says android device only for G4 without share. I tried android NS with G5 and got nothing. Now I'm back to G4 temporarily but switched to apple for G5. Is NS available on the desktop computer? That graph reveals that insulin needs vary over time. How do you use the graph?
Ben West
@bewest
Jan 22 2016 05:50
@amazaheri nice, make sure to add --enable-experimental to the configure step
the graph shows all the pump history and cgm on single display
Ben West
@bewest
Jan 22 2016 06:09
nice
find /var/cache/pbuilder/raspbian-jessie-armhf/result/
/var/cache/pbuilder/raspbian-jessie-armhf/result/
/var/cache/pbuilder/raspbian-jessie-armhf/result/bluez-cups_5.36-1~exp1rpi1_armhf.deb
/var/cache/pbuilder/raspbian-jessie-armhf/result/bluez_5.36-1~exp1rpi1_armhf.deb
/var/cache/pbuilder/raspbian-jessie-armhf/result/bluez_5.36-1~exp1rpi1.tar.gz
/var/cache/pbuilder/raspbian-jessie-armhf/result/bluez-obexd_5.36-1~exp1rpi1_armhf.deb
/var/cache/pbuilder/raspbian-jessie-armhf/result/libbluetooth-dev_5.36-1~exp1rpi1_armhf.deb
/var/cache/pbuilder/raspbian-jessie-armhf/result/bluez-dbg_5.36-1~exp1rpi1_armhf.deb
/var/cache/pbuilder/raspbian-jessie-armhf/result/bluetooth_5.36-1~exp1rpi1_all.deb
/var/cache/pbuilder/raspbian-jessie-armhf/result/libbluetooth3_5.36-1~exp1rpi1_armhf.deb
/var/cache/pbuilder/raspbian-jessie-armhf/result/bluez-hcidump_5.36-1~exp1rpi1_armhf.deb
/var/cache/pbuilder/raspbian-jessie-armhf/result/bluez-test-scripts_5.36-1~exp1rpi1_all.deb
/var/cache/pbuilder/raspbian-jessie-armhf/result/libbluetooth3-dbg_5.36-1~exp1rpi1_armhf.deb
/var/cache/pbuilder/raspbian-jessie-armhf/result/bluez_5.36-1~exp1rpi1.dsc
/var/cache/pbuilder/raspbian-jessie-armhf/result/bluez_5.36-1~exp1rpi1_armhf.changes
bewest@bewest-MacBookPro:~/Downloads/bluez-5.36$
maybe compiled deb for working bluetooth on jessie
Dana Lewis
@danamlewis
Jan 22 2016 06:14
@aaybob you can ssh into the Pi from your phone using an app like server auditor (free). Just need the Pi to be online from wifi or your phones hotspot. And you absolutely need to view the logs before you run this without watching (I.e. Overnight)
You can do Nightscout with the G4, with or without share. You can view Nighrscout from a desktop or by phone.
Highly recommend you set it up and don't operate the loop unless you have a view (logs or NS) into what the system is doing. Otherwise, you can't trouble shoot and understand what the system is doing, and you also can't decide effectively whether you need to take action or not beyond what the loop is doing
Dana Lewis
@danamlewis
Jan 22 2016 06:21
Regarding the above ^^, I would appreciate suggestions (here, or via a PR to the docs) for those working through the process on how to make this more clear in the docs re: steps for logs and importance of them.
Ali Mazaheri
@amazaheri
Jan 22 2016 06:22
@bewest just install 5.33 and seems to be working fine with --experimental
Ben West
@bewest
Jan 22 2016 06:22
awesome
on edison?
yeah 5.33 should work I believe
looks like anything around 30-ish
their release notes mention cleanup of gattt handling
Ali Mazaheri
@amazaheri
Jan 22 2016 06:23
yeap
Ben West
@bewest
Jan 22 2016 06:23
on bright side, I think I just compiled working debian for jessie
Ali Mazaheri
@amazaheri
Jan 22 2016 06:23
edison@jarvis:~$ ps aux | grep bluetoothd root 1906 0.0 0.2 5176 1996 ? S 22:21 0:00 bluetoothd --experimental edison 1971 0.0 0.0 4300 904 pts/0 S+ 22:22 0:00 grep bluetoothd edison@jarvis:~$
Ben West
@bewest
Jan 22 2016 06:23
edison is special, not sure this build method will work
Ali Mazaheri
@amazaheri
Jan 22 2016 06:23
:+1:
Ben West
@bewest
Jan 22 2016 06:24
nice
and openaps use share iter_glucose 2 works?
Ali Mazaheri
@amazaheri
Jan 22 2016 06:24
that's next
Aaron Michelson
@Aaybob
Jan 22 2016 06:38
Okay but... the logs tell you the algorithm is being followed. The details of when it shuts basal off and turns it back on during minor fluctuations doesn't seem informative. Sure at night it can't be check real time. After the fact the CGM graph is pretty informative. I'm running the crontab to update on fifteen minute increments so checking the activity realtime on the pump seems perhaps more useful than the recorded logs. And by useful I mean easier to understand. I think that 15 minute will spare battery life and the first graph I saw was from a first time AP'er who had the thing working to level his BG readings but toggling the basal rapidly. I'm not wearing it at night so you can be assured. I want to watch what it is doing. But that night of good glucose readings made a difference healthwise.
Ben West
@bewest
Jan 22 2016 06:40
wheezy and jessie debs both baked :-D
not sure where to put them
Dana Lewis
@danamlewis
Jan 22 2016 06:44
@aaybob that might have been my graph, heh. I guess I am concerned having seen a post that said something like "I don't know what the system did". If something was non-optimal and you do get an unexpected/less-optimal result, having the logs helps us help you troubleshoot to see if you found an edge case in the code we need to fix, or if you need to tweak your implementation.
Thomas Emge
@ThomasEmge
Jan 22 2016 07:09
@amazaheri so you have a TI stick working on edison?
Oskar Pearson
@oskarpearson
Jan 22 2016 08:11
Thomas Emge
@ThomasEmge
Jan 22 2016 08:12
do you have the dexcom share running over bluetooth as well?
Oskar Pearson
@oskarpearson
Jan 22 2016 08:13
These days the subg rfspy firmware works with the ti stick too - and it's a better option overall
Siobhan doesn't use dexcom - not sure if anyone else is though
Thomas Emge
@ThomasEmge
Jan 22 2016 08:18
Thanks, as soon as someone has a working combination with the dexcom I do see a edison in my future...and I find to find a cc debugger...
Dana Lewis
@danamlewis
Jan 22 2016 08:19
@ThomasEmge That might be us this weekend if all goes well :)
Thomas Emge
@ThomasEmge
Jan 22 2016 08:23
ohh, that would be nice :)
Thomas Emge
@ThomasEmge
Jan 22 2016 08:35
@danamlewis I saw that Ben compiled something ...I myself haven't been able to get it going on either wheezy or jessie :(...looking forward to your findings
Ben West
@bewest
Jan 22 2016 08:58
the thing I compiled is bluez for pi
Daniel Bjørnbakk
@danibjor
Jan 22 2016 11:29
Anyone know if it's possible to talk to the Medtronic CareLink USB -stick from Windows 10?
Or maybe get access to the raw usb port via vendor/product id
Oskar Pearson
@oskarpearson
Jan 22 2016 11:42
@danibjor Don't know for sure. But if it's on windows 10 I'm guessing you might find it has USB 3.0 ports. And USB 3 causes lots of issues for Carelink devices as far as I know.
@bewest Edison should be the same, I hope, since it has the same base OS in our case. Have you got instructions somewhere? I'd like to give it a try
Daniel Bjørnbakk
@danibjor
Jan 22 2016 11:44
@oskarpearson still researching
Windows 10 sample code for usb access
using WinUSB.sys driver
I see a usage both on my Lumia 950 and on the PI running Win10 IOT
The Lumia 950 have USB host capabilities, and if I could talk to the CareLink via a custom driver/layer in C# in a Universal App, the code could run on the Phone and no extra hardware/boxes needed
Oskar Pearson
@oskarpearson
Jan 22 2016 11:50
Would you need to re-write the whole of openaps and decocare in c# though?
Daniel Bjørnbakk
@danibjor
Jan 22 2016 11:54
Probably, but Microsoft working on Docker support on Windows IOT - so there might be a shortcut to reuse existing code
There are SDKs for Universal Apps that lets you create drivers to embedd in the application, but those have to be written in C, witch is out of my league
Those work on both desktop, mobile and universL
universal *
Rachel Sandlain
@audiefile
Jan 22 2016 15:30
@danibjor @amazaheri has done work on using openaps on Windows. I'd reach out so you don't have to re event the wheel
Daniel Bjørnbakk
@danibjor
Jan 22 2016 15:33
sweet :)
Trying to get the CareLink stick working on Windows 10 universal apps via the WinUSB generic driver that comes with Windows 10. If that works, then it should work on both desktop, mobile, IOT (RPI) - and on the Xbox One, tough that one would be a bitch to carry around
Alin Artiom Kenibasov
@ktomy
Jan 22 2016 17:01
Is there somewhere a list of write-enabled firmware versions?
i.e. for640 v2.6D is it possible to enact something?
Dana Lewis
@danamlewis
Jan 22 2016 17:02
Pumps? The hardware page in docs is the best we've got
Ali Mazaheri
@amazaheri
Jan 22 2016 17:02
Capture.PNG
@bewest ^^
:)
Paul Andrel
@stavlor
Jan 22 2016 17:03
@amazaheri :+1:
Alin Artiom Kenibasov
@ktomy
Jan 22 2016 17:05
@danamlewis thank you , 640 is not listed :) so it's a bad pump :)
Scott Leibrand
@scottleibrand
Jan 22 2016 17:11
it's pretty good as a pump. not so good for OpenAPS. :-)
Matthias
@ceben80
Jan 22 2016 17:11
Unfortunately the Animas Vibe is not listed too :worried: would be so nice because the pump has the CGM receiver onboard
Scott Leibrand
@scottleibrand
Jan 22 2016 17:11
do you know of any way to remote control the Vibe? IIUIC it doesn't have a remote like the Ping does.
Matthias
@ceben80
Jan 22 2016 17:12
the Vibe has an IR interface....
Luckily I still have an old 722 for looping purpose,but the Vibe would be so great
Jason Calabrese
@jasoncalabrese
Jan 22 2016 17:15
think the ir only works when the pump is suspended like the ping
Milos Kozak
@MilosKozak
Jan 22 2016 17:16
IR will be probably read only on vibe
Matthias
@ceben80
Jan 22 2016 17:16
Yes that's true
@MilosKozak This is really sad
dougtakeuti
@dougtakeuti
Jan 22 2016 17:34
Hello bunch of smart people! Hope everyone is fine in whatever time zone you are. I'm trying to set up the uploading to NS. As I review the documentation, some doubts are cropping up:
1 - the "zoned" term for clock-zoned.json and pump-history-zoned.json is there just to emphasize the requirement for everything to be in the same TZ or there is some additional processing called for?
2 - When upload-status. json is invoked, predict/oref0.json is required. Is this the outcome of determine-basal?
3 - Under status-upload, is devicestatus.json the same as current_pump_status?
Scott Leibrand
@scottleibrand
Jan 22 2016 17:40
devicestatus.json is a NS API endpoint, not a local file on the Pi
determine-basal enforces that the time on the CGM matches the time on the Pi (unrelated to uploads, but it won't enact anything otherwise)
I don't know that we enforce the timezone on the pump: I think we just check pumphistory against clock for calculating IOB
but if your timezones are wrong, your temp basals will show up incorrectly in NS
I'm not sure where you saw predict/oref0.json: I don't have that.
Ali Mazaheri
@amazaheri
Jan 22 2016 18:33
@ThomasEmge sorry for late response. Yes to your question and also running Dex thru BLE even though we are using G5
This message was deleted
Ali Mazaheri
@amazaheri
Jan 22 2016 18:41
This message was deleted
@danibjor Let me know if you want to get more detail, we ported the openaps to Windows 10 with changes to decoding-carelink and Dexcom and I have the solution end to end running. We as a team also talked about a Mobile app being cross platform etc. At this point BLE connectivity to RL is key which we already have that for iOS and if we want to go that route I am inclined to start with iOS first. Assuming we can hash out the BLE stack on Windows with RL I still see porting all the python code to C# an huge effort along with all C++ code required to make it happen. My 2 cents.
Thank you @audiefile for mention
Rachel Sandlain
@audiefile
Jan 22 2016 18:52
Your welcome :smile:
Thomas Emge
@ThomasEmge
Jan 22 2016 19:24
@amazaheri congrats on getting the BLE connection working. I would like to stay with the G4 a little longer as the raw values are really nice in NS so I am looking for a BLE share receiver connection. On the topic of TI stick: do you have access to a cc-debugger? I believe you are somewhat close to my geographical area so I am looking for some assistance if possible wink wink.
Jim Matheson
@jmatheson
Jan 22 2016 19:41
is anyone using a USB hotspot (as opposed to a wifi card with tethering)?
Ali Mazaheri
@amazaheri
Jan 22 2016 20:25
I do have a CC, where are u located>
Thomas Emge
@ThomasEmge
Jan 22 2016 20:46
@amazaheri Redlands, in the inland empire. However I am down around the Irvine/Costa Mesa area on a regular basis as my son has soccer practice down there.
Dragan Maksimovic
@dm61
Jan 22 2016 21:14
Hi @juliatakeuti, I've had similar questions when I was setting up ns upload recently. I've added some more details to the visualization section, which you can see here: https://github.com/dm61/docs/blob/dev/docs/Automate-system/vizualization.md.
Ali Mazaheri
@amazaheri
Jan 22 2016 21:50
@ThomasEmge
great let me knlw when and we can plan it
Thomas Emge
@ThomasEmge
Jan 22 2016 21:53
@amazaheri Awesome, thanks.
Alin Artiom Kenibasov
@ktomy
Jan 22 2016 23:01
we should handle this in a way:
blob
after callibration delta is completely wrong
openaps/oref0#59
Alin Artiom Kenibasov
@ktomy
Jan 22 2016 23:10
Ideas are wellcomed :) will try to implement a solutino after we'll find out what actions should be done.
Scott Leibrand
@scottleibrand
Jan 22 2016 23:10
avgdelta isn't usually too far off...
obviously if we pull calibration records we could ignore data points prior to calibrations
but I actually haven't seen any bad actions resulting from calibration deltas
Alin Artiom Kenibasov
@ktomy
Jan 22 2016 23:13
didnt' cought a screenshot, but it changed basal from 0.3 to 1.15
Scott Leibrand
@scottleibrand
Jan 22 2016 23:14
it will usually overreact a little bit, but in the correct direction, so when it cancels 15m later it doesn't have to reverse course at all
you can scroll back in NS
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:18
I've seen that before, it's annoying, but probably only an issue if the pump goes out of range
but if we knew when calibrations happened it would be good to ignore everything before the calibration
Alin Artiom Kenibasov
@ktomy
Jan 22 2016 23:19
hm, just a moment, let me understand something... it doesn't want to stop temp... just a second
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:21
I have a feeling that this has become more of an issue for us after switching from one touch strips to contour, my gut feeling is that the low bias of the one touch aligned more with the dexcom, and that somehow dex is fighting the contour
Alin Artiom Kenibasov
@ktomy
Jan 22 2016 23:21
so after callibration
blob
and now
blob
just a second, i'll see determine-basal input
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:25
calibration/dex accuracy issues screwed us up today, and on Tuesday
Alin Artiom Kenibasov
@ktomy
Jan 22 2016 23:25
ah, ok, it cancelled the temp basal but I didn;; see it :)
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:25
today was bad, set off a chain of events, will be hours to fully recover from it
Alin Artiom Kenibasov
@ktomy
Jan 22 2016 23:26
@jasoncalabrese I don't know how it is for NS, can I open an issue with a wanted feature, or it is like "you want it - you write it"? actually what I wnt is to see basal value on these blue arrays
now I can see only temp-basal values
Dana Lewis
@danamlewis
Jan 22 2016 23:27
@jasoncalabrese when are ya'll calibrating? Are you consistent?
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:27
calibration was never, never an issue for us
always calibrated almost every time
Milos Kozak
@MilosKozak
Jan 22 2016 23:28
@ktomy you se basal in a pill ... if you go to retro mode you see basal from retro time too
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:28
really wondering if it was that the bias of the dex and one touch aligning that helped
Alin Artiom Kenibasov
@ktomy
Jan 22 2016 23:28
I know, it is just a better visualisation mode, at least for me.
Dana Lewis
@danamlewis
Jan 22 2016 23:28
Just wondering if it's worth continuing with that meter, or otherwise changing calibration pattern
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:29
or maybe it's a bad sensor
so many factors
Dana Lewis
@danamlewis
Jan 22 2016 23:29
Yea
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:29
I changed almost everything over a pretty short period
Dana Lewis
@danamlewis
Jan 22 2016 23:30
Remind me, did you say you're doing side by side comparison to old meter?
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:30
I did for a day
contour was always higher
the integration is pretty nice, don't want to give that up
any finger stick gets uploaded to NS automatically now
so makes using the bolus wizard easy
Dana Lewis
@danamlewis
Jan 22 2016 23:33
Wondering if calibrating less often would help.
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:34
but if it's off and you don't calibrate it seems you're worse off
Dana Lewis
@danamlewis
Jan 22 2016 23:34
How off is off? 10, 20, 30, more?
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:35
was 96 going into morning snack, check was 70
so loop wasn't trying to help much before, then they over corrected
calibration at the top of high following that was cgm 204 vs meter 242
but other calibrations after have been close
hasn't been that high in a long time
I call it sabotage, damn humans
Dana Lewis
@danamlewis
Jan 22 2016 23:38
We humans are the worst
Ali Mazaheri
@amazaheri
Jan 22 2016 23:39
I see the same issue as well, there are times that loop keeps everything steady and cgm showing in 90's and meter is in 140s
not happening very often
common issue with any closed loop unless something magical happens with CGM sensors :) to be accurate 150% of times
G10?
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:42
wonder what the commercial APs will do, I'm expecting them to give up quickly like dex does with ???
Ali Mazaheri
@amazaheri
Jan 22 2016 23:43
honestly they have the same issue
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:44
I bet they will be overly cautious
Ali Mazaheri
@amazaheri
Jan 22 2016 23:44
BLACKBOX ALERT!
Dana Lewis
@danamlewis
Jan 22 2016 23:44
Yes.
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:45
yup
Ali Mazaheri
@amazaheri
Jan 22 2016 23:45
not like @oskarpearson black box for his implementation though :)
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:45
seems like we have the data (at least with the g4) to do something smarter
Ali Mazaheri
@amazaheri
Jan 22 2016 23:46
I don't :(
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:47
maybe they'll need to open things up for APs with the g5x or whatever it will be called
I think if we can get calibrations we should ignore previous bg's for calculating delta
Alin Artiom Kenibasov
@ktomy
Jan 22 2016 23:52
it's my opinion too. Of course it is not "prio1" issue, but this is on wish-list.
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:53
maybe a filter step in our own loops
Kevin Wells
@wkcwells
Jan 22 2016 23:53
I find that two calibrations in a row can be as far apart from each other as a calibration from the dex. I.e. the calibrations themselves are not always more accurate than a well calibrated dex.
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:53
then nothing else changes
Kevin Wells
@wkcwells
Jan 22 2016 23:53
Does anyone else have that experience?
Dana Lewis
@danamlewis
Jan 22 2016 23:54
@wkcwells You mean two finger sticks in a row?
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:54
so far contour does seem to be closer for the 2 startup drops then one touch
Dana Lewis
@danamlewis
Jan 22 2016 23:56
I calibrate 2x day, am and bedtime. If I feel like my CGM is off, I'll test and do an additional calibration if it's 15-20+ off. But otherwise, they're not same point in time measurements so don't expect them to align perfectly
Kevin Wells
@wkcwells
Jan 22 2016 23:56
@danamlewis Yes. When I got started with CGM (including MinMed) I used to cross check everythign a lot more. (Because I needed to because the sensor was not accurate.)
Dana Lewis
@danamlewis
Jan 22 2016 23:56
Yea. Meter can have a ton of variability, too, of course, which is the other reason I don't calibrate a lot more
Kevin Wells
@wkcwells
Jan 22 2016 23:56
So this is somewhat out of date experience, but it was much more often that the double stick at the beginning of run.
Meaning I would double stick much more often.
Alin Artiom Kenibasov
@ktomy
Jan 22 2016 23:57
@danamlewis the second calibration immediately or after 10-15 minutes?
Dana Lewis
@danamlewis
Jan 22 2016 23:58
For startup, I do the two fingersticks and enter at the same time.
Kevin Wells
@wkcwells
Jan 22 2016 23:58
I believe I’ve been told that the reason that the dex does not jump all the way to a calibration that is different is that their calibration algorithm tries to account for finger stick inaccuracy.
My double sticks were immediate.
Alin Artiom Kenibasov
@ktomy
Jan 22 2016 23:59
and for this situation: "I'll test and do an additional calibration if it's 15-20+ off" ?
Jason Calabrese
@jasoncalabrese
Jan 22 2016 23:59
I heard is was a patent issue, that MM has a 1 drop calibration patent