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

15th
Apr 2018
Martin Haeberli
@mhaeberli
Apr 15 2018 00:03
wondering whether we can use ML on nightscout data and/or openhumans data to learn how to set DIA and other openaps parameters per patient...
(also wondering why ‘crickets’ on this channel since this time yesterday)
Scott Leibrand
@scottleibrand
Apr 15 2018 00:32
Because all the troubleshooting has been misdirected to oref0. :smile:
Which, incidentally, is where ML/algo convos usually end up. ;-)
We did come up with an idea for gradient descent method for estimating DIA with autotune, but never implemented it.
Martin Haeberli
@mhaeberli
Apr 15 2018 00:38
cool ...
so I should switch to oref0? or will traffic resume over here?
do you have any notes on the gradient descent DIA?
smb2.png
just a note of thanks! SMB more or less fully on now. normal insulin, NOT Fiasp. carbs announcements only. very happy college user !
you also see the last ‘bump’ - here a case where -maybe- we know more about DIA, albeit we don’t know WHAT carbs were declared...
Scott Leibrand
@scottleibrand
Apr 15 2018 01:57
Normally this channel is for troubleshooting and oref0 is for algorithm stuff. So sure, let’s move the algo discussion there.
Martin Haeberli
@mhaeberli
Apr 15 2018 02:58
K thx
David Rimmer
@zjedr
Apr 15 2018 06:25
Has anyone got solutions to these problems when running fully offline?
  1. On rig boot datetime defaults to 01/01/2000. Without ntp server access datetime has to be set manually, however if pump datetime <> rig datetime it wont be updated as that section is bypassed if no ntp.
  2. How do you change pump serial number and/or phone BT mac when oref-setup can't be run offline?
Andrew Kirkby
@kirkbysound
Apr 15 2018 13:04
I’ve a quick query, my tethering is being a bit sporadic (probably more of an iPhone thing than an openaps one) from what I understand part of the crontab takes care of internet kick starting tethering when required. Could someone point me in the direction of the line which runs the script please? I’d like to add a few bits to the script that basically reboots the rig after 30 minutes of no connectivity. Much appreciated!
James Babcock
@jimrandomh
Apr 15 2018 13:22
@kirkbysound The script you're looking for is oref0-online.sh, and it's run as a cronjob
Andrew Kirkby
@kirkbysound
Apr 15 2018 13:40
I’ll take a look, thanks!
Scott Leibrand
@scottleibrand
Apr 15 2018 16:18
@zjedr you can change the pump serial and BT MAC manually in the appropriate ini files and in crontab -e, and then of course pair the new MAC in bluetoothctl as you normally would.
Or connect to wifi and do it over ssh.
If you’d like to help troubleshoot why the rig isn’t getting time from the pump while offline, I think there is a regression there we need to fix.
Martin Haeberli
@mhaeberli
Apr 15 2018 17:33
20180415Eden1.png
so the question was - why were numbers not coming down around 1 am. Theorized that set was at 72+ hours so changed it. but maybe there is other magic we could do?
user1109
@user1109
Apr 15 2018 17:50
i need help. I can loop only for 6-8 hours. logrotate is not working. I changed from daily to hourly, used apt-get clean and did all discussed here: https://gitter.im/nightscout/intend-to-bolus?at=5acd42db7c3a01610dcc1192. Nothing helped me
Martin Haeberli
@mhaeberli
Apr 15 2018 17:53
@user1109 - what does df -h show?
Martin Haeberli
@mhaeberli
Apr 15 2018 18:23
(that tells you how much free space there is)
Tim Gunn
@Tornado-Tim
Apr 15 2018 20:09
Is it usual that the pump reservoir amount is incorrect? I have a low cartridge but OpenAPS/NS think i have 0 units left.....
katie disimone
@Kdisimone
Apr 15 2018 20:24
I have about 6-7 Edison/explorer rigs to sell. Flashed with jubilinux. $160 plus shipping. And a single Edison flashed with jubilinux (no explorer board) $80 plus shipping. Send me a message if you are interested. Thanks!
Scott Leibrand
@scottleibrand
Apr 15 2018 21:56
@Tornado-Tim how many units does the pump say it has left?
Thomas
@TH2100
Apr 15 2018 22:09
My oref0 0.6.1 with a 512 medtronic uses quite bad ISFs, which I just can influence with temp targets. The pumps ISFs aren’t read nore are nightscout-profiles. Is there a pref-file on openaps I could edit for ISF changes?
Scott Leibrand
@scottleibrand
Apr 15 2018 22:24
@rollerdecom you’re not ready to closed loop yet if you don’t have the correct ISF set in your rig. Please go back and re-review the docs and find the place you’re supposed to set all of the settings in the rig that can’t be read from a 512.
Thomas
@TH2100
Apr 15 2018 22:28
there are basal_profile.json, settings.json, bg_targets_raw.json to be edited for 512: no isf. then there is edit-pref: no isf
Scott Leibrand
@scottleibrand
Apr 15 2018 22:31
It’s called sensitivity, not ISF.
Thomas
@TH2100
Apr 15 2018 22:39
Sensitivity ratio is 1. ISF on the pump is 26, latest autotune said 30 and openaps works with 40. So I just wonder what to change. Of course, I’m missing something. And as searching the docu didn’t help and as this was a problem I’m having for weeks now…
Scott Leibrand
@scottleibrand
Apr 15 2018 23:21
This message was deleted
K. What does pump-loop.log say about ISF?
Also run: grep sensitiv ~/myopenaps/settings/*
Tim Gunn
@Tornado-Tim
Apr 15 2018 23:44
@scottleibrand it says ---.-- U left
Time left 0
Got no clue why
Scott Leibrand
@scottleibrand
Apr 15 2018 23:47
The pump considers the reservoir “empty” when it has about 6-8U left.
Tim Gunn
@Tornado-Tim
Apr 15 2018 23:48
ok so thats normal?
James Babcock
@jimrandomh
Apr 15 2018 23:48
Right, the pump believes it has 0, but continues to try tries squeezing on the cartridge to give you a little bit of grace period, and there's a little more past the 0 mark but you shouldn't rely on that
If you don't refill it soon, you'll stop receiving insulin and get a no-delivery alarm on the pump, not necessarily in that order
Tim Gunn
@Tornado-Tim
Apr 15 2018 23:49
yea thats fine, I am just used to running my Animas pump before this pump right to empty, to avoid waste
James Babcock
@jimrandomh
Apr 15 2018 23:50
That's not really recommended. A better but still not-recommend-maybe-risky way to avoid wasting the last bit of insulin is to refill the same cartridge, rather than start from an empty cartridge.
Tim Gunn
@Tornado-Tim
Apr 15 2018 23:51
because the res status on the pump said it wasnt empty (had one bar left), when OpenAPS said it had 5 U left, which seems rather odd
the two scales are not equal, and that is rather confusing