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

30th
Mar 2016
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 00:00
ah cra
p
somehow i am back on the dev branch.
no idea how that happened
very strange
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 00:07
ok back on AMA. still not working though
Scott Leibrand
@scottleibrand
Mar 30 2016 00:07
what is the error?
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 00:07
no error, output file is just blank.
and all input files on the surface look valid
Scott Leibrand
@scottleibrand
Mar 30 2016 00:08
no output to the terminal at all?
you might need to run npm install
I usually do cd && test -d src/oref0 && cd src/oref0 && npm install && sudo npm install -g && sudo npm link && sudo npm link oref0
not sure which of those are actually required
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 00:10
ok will try that
Scott Leibrand
@scottleibrand
Mar 30 2016 00:19
if that doesn't work, run the oref0-detect-sensitivity script manually (not via openaps) and make sure that works.
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 00:22
alright seems like it is actually running now. thanks. will see what it dumps out.
says they are in normal range.
i thought it used to dump out more info
Scott Leibrand
@scottleibrand
Mar 30 2016 00:25
it did before I refactored it to use the same library for COB
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 00:25
ok
Scott Leibrand
@scottleibrand
Mar 30 2016 00:25
you can re-enable the console.error statements if you want
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 00:26
k
also, i think you mentioned on AMA it would push a temp basal all the time and you could see when loop was going in NS from that. Is that the case or do i actually need to rework my alias for enact
Scott Leibrand
@scottleibrand
Mar 30 2016 00:29
yeah, it should always recommend a temp, so your existing enact should run that
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 00:30
ok, will monitor. should be awesome. really happy i’ve got my own ini now where i understand all of my aliases inside and out
moved pulling some of the reports into the batch at the end of the loop and only pulling the big batch of history once every 60 mins
JaysonEwer
@JaysonEwer
Mar 30 2016 00:33
That sound sweet, @CrushingT1D nice job.
dougtakeuti
@dougtakeuti
Mar 30 2016 00:51
Hey there, I've been out of the loop for a few days, but I remember that on Saturday @scottleibrand mentioned the possibility of getting 522 pumps here in Brazil, something that I had told @bewest and him during a fb chat a few months ago. Well, I'm sorry to say that I got the official word from my Medtronic rep that those are no longer sold by them. :disappointed:
Dana Lewis
@danamlewis
Mar 30 2016 00:51
:( :(
Ben West
@bewest
Mar 30 2016 01:15
just added new --cli switch to report show --cli:
$ openaps report show oref0-predict/oref0.json --cli
openaps use determine-basal shell oref0-monitor/iob.json monitor/temp-basal-status.json monitor/glucose.json oref0-monitor/profile.json []
now, if you're report won't run and you're scratching your head to figure out how to run the use for it manually, this thing will just tell you what would run under the hood
dougtakeuti
@dougtakeuti
Mar 30 2016 01:19
:+1:
Jason Calabrese
@jasoncalabrese
Mar 30 2016 01:20
thats nice
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 01:21
:+1:
Scott Leibrand
@scottleibrand
Mar 30 2016 01:24
Awesome. That will be really useful.
Eric Luhrs
@eluhrs
Mar 30 2016 01:28
Where's the best place to ask about xdrip hardware? I want to make a receiverless setup, but I can't find any compatible hm10-based bluetooth modules. Wondering if I'm just missing them or if there are any other options.
Dana Lewis
@danamlewis
Mar 30 2016 01:28
there’s a nightscout/dexdrip channel here
Eric Luhrs
@eluhrs
Mar 30 2016 01:29
Thanks, @danamlewis . I was searching for xdrip....
Dana Lewis
@danamlewis
Mar 30 2016 01:29
:)
Jason Calabrese
@jasoncalabrese
Mar 30 2016 01:34
don't see how cob can be right, the old cob that we have in NS is showing 4g, where oref0 has 28g
even with the 28h history
Scott Leibrand
@scottleibrand
Mar 30 2016 01:36
The NS COB is definitely too low. Always has been, except in the wip/iob-cob branch.
Has BG risen as much as would be predicted by your carb absorption rate, minus the IOB decay?
Most likely not, so AMA is still expecting those carbs to maybe hit at some point. It won't give up on them until 1.5*DIA.
But if they're really gone, and BG deviation never rises, the COB have no effect.
Jason Calabrese
@jasoncalabrese
Mar 30 2016 01:41
if you get the timing right and there isn't enough of a rise they won't decay at all till 1.5*DIA?
Scott Leibrand
@scottleibrand
Mar 30 2016 01:42
No, you need to understand deviation.
Jason Calabrese
@jasoncalabrese
Mar 30 2016 01:42
that about what it looks like, could make it really simple and just add up all the carbs for dia * 1.5 in that case
Scott Leibrand
@scottleibrand
Mar 30 2016 01:42
Deviation is the amount BG is rising over and above BGI
Which is how much BG should be falling from insulin activity.
So if you have enough insulin active that you should be dropping 10mg/dL every 5m, but you're actually +0, that is a deviation of 2mg/dL/min
That, in turn, lets you calculate how fast carbs are absorbing, based on your carb ratio.
and ISF.
What are Andrew's ratios?
Jason Calabrese
@jasoncalabrese
Mar 30 2016 01:47
I get how it should work, but what I see is carbs always being active till and then to fall off the limit
63 to 28 in as soon as the time limit is up, and then 28 till that limit is up
Scott Leibrand
@scottleibrand
Mar 30 2016 01:48
Are you entering the carbs all at once, or stacked?
I wonder if your carb ratio is correct.
Jason Calabrese
@jasoncalabrese
Mar 30 2016 01:49
mostly all at once before a snack/meal
if anything it's higher than actual
at 16 most of the day
Scott Leibrand
@scottleibrand
Mar 30 2016 01:55
Higher than actual would cause this behavior.
Hmm, maybe not.
16 means 16g per 1U, which means ISF/16
Scott Leibrand
@scottleibrand
Mar 30 2016 01:57
So 80/16 mg/dL per g, right?
Jason Calabrese
@jasoncalabrese
Mar 30 2016 01:57
80 might be low, bet it's closer to 90
yeah
so 1g we'd expect a rise of 5 mgdl?
Scott Leibrand
@scottleibrand
Mar 30 2016 01:59
So an easier way to analyze this is look at starting BG, total bolus+net basal insulin, and ending BG after DIA
to see if the ratios are right.
Correct
and carbs of course.
Is his NS URL public?
Jason Calabrese
@jasoncalabrese
Mar 30 2016 02:01
I'll give it to you, I used to have a mirror, but the mongo dump/restore trick isn't working will all this data
Scott Leibrand
@scottleibrand
Mar 30 2016 02:02
which meal should we look at?
what's the one where you saw COB stay at 28 until 1.5*DIA?
Jason Calabrese
@jasoncalabrese
Mar 30 2016 02:06
right now it's showing 28g, but doesn't know about the most recent 4g
Dana Lewis
@danamlewis
Mar 30 2016 02:06
Just submitted the big docs re-structure as a PR to dev - take a look, provide feedback and thumbs up when you have time in the next day or so! openaps/docs#114
Jason Calabrese
@jasoncalabrese
Mar 30 2016 02:07
so it's using the 10g correction from 3pm and the 18g snack at 5:30
with no decay
Screenshot 2016-03-29 19.08.10.png
if anyone else is curious
I thought they had corrected the low at school, but they didn't see/hear it
Scott Leibrand
@scottleibrand
Mar 30 2016 02:09
why is meal.json an hour old?
that has to be updated every loop
Jason Calabrese
@jasoncalabrese
Mar 30 2016 02:10
it's just not changing
like autosens.json
Scott Leibrand
@scottleibrand
Mar 30 2016 02:11
can you run oref0-meal manually?
your deviation is way positive, so it should be updating
you might not be refreshing one of its inputs or something? not sure.
Jason Calabrese
@jasoncalabrese
Mar 30 2016 02:13
$ openaps report invoke monitor/meal.json && cat monitor/meal.json
meal://text/shell/monitor/meal.json
.......
Sensitivity within normal ranges
reporting monitor/meal.json
{"carbs":28,"boluses":1.1,"mealCOB":28}
Scott Leibrand
@scottleibrand
Mar 30 2016 02:14
oh, I found it
your glucose.json is a day old
Jason Calabrese
@jasoncalabrese
Mar 30 2016 02:15
it doesn't get pushed
Jason Calabrese
@jasoncalabrese
Mar 30 2016 02:15
with the deletes
Scott Leibrand
@scottleibrand
Mar 30 2016 02:16
glucose-ns.json is up to date, but not glucose.json, which is what that report uses
Jason Calabrese
@jasoncalabrese
Mar 30 2016 02:16
it's current on disk
Scott Leibrand
@scottleibrand
Mar 30 2016 02:16
hmm k
Jason Calabrese
@jasoncalabrese
Mar 30 2016 02:16
with the deletes it doesn't get git added
1 of the many little things that bugs me
added and pushed
Scott Leibrand
@scottleibrand
Mar 30 2016 02:17
k
now it's just missing
Jason Calabrese
@jasoncalabrese
Mar 30 2016 02:18
yeah, saw that, annoying
happens to everything that isn't run as a report
Scott Leibrand
@scottleibrand
Mar 30 2016 02:20
I would make copies of all the inputs, make sure they're not empty, and run oref0-meal manually (not as an openaps use/report)
if that still shows 28, it's time to start enabling debug print statements
Jason Calabrese
@jasoncalabrese
Mar 30 2016 02:22
hoping to write a little test for it tonight
Scott Leibrand
@scottleibrand
Mar 30 2016 02:23
that would also be good
FWIW, there is a known issue with carb stacking, where it will do a simplified calculation, assuming all of the carbs started decaying at the time of the larger carb entry. that can give some weird results, but it won't stop things decaying entirely like this.
Jason Calabrese
@jasoncalabrese
Mar 30 2016 02:26
damn, busy looking at that and I forgot to bolus for dinner, and BG was +30 from dex
change of plans
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 05:09
:(
sorry to hear that
i’d love to have an app on my watch that would alert me if i am putting food in my mouth and have not bolused :)
seems like it would be possible to do, well if i ate with my left hand.
Matthias Granberry
@mgranberry
Mar 30 2016 05:11
@CrushingT1D I was joking with one of Google's activity recognition folks about that and she seemed quite interested.
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 05:12
it seems easy enough to track movement
the main issue is i am american so i eat with my right hand. So either i need to wear a watch there or do full Australia conversion if/when I move and go left hand all the time
Matthias Granberry
@mgranberry
Mar 30 2016 05:13
She had just finished a push-up/jumping jack recognizer. She didn't think it would be too difficult to train
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 05:13
think it could detect the difference between eating and drinking?
Matthias Granberry
@mgranberry
Mar 30 2016 05:14
who knows, but I sure don't have a tagged data set of a few hundred people. Google does, though, from dogfooding and place detection in maps.
it would be pretty easy for them to do.
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 05:17
location might work, but arm movement would be most accurate.
Matthias Granberry
@mgranberry
Mar 30 2016 05:18
I was only thinking about how to narrow down the data collection.
Location works great for correlating boluses with food. I have some scripts that do it for me.
Jason Calabrese
@jasoncalabrese
Mar 30 2016 05:30
trying 2 edisons with TI sticks now, 1 has a 30s preflight sleep, the other has the normal 2m preflight sleep
I could see the rsync and quick enact working good in this setup
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 05:31
is quick enact only useful for “falling fast”?
i have it planned to add, but i dont have that problem often enough for it to be a priority for me
Jason Calabrese
@jasoncalabrese
Mar 30 2016 05:32
I think that was the initial use for it
but I was thinking that if you're syncing the json files between the rigs, 1 could enact via the history, settings, etc of the others
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 05:34
:minibus:
there is no mind blown emoji so i went with the “I am on that mini bus"
Jason Calabrese
@jasoncalabrese
Mar 30 2016 05:35
Screenshot 2016-03-29 22.34.19.png
@scottleibrand here's an example I going to try to setup a test for, the 36g was dinner, at 7:25, now 3hrs later after some rise COB is still 36g
we'd expect some decay in this case right?
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 05:41
is that from NS or pump entered carbs?
Jason Calabrese
@jasoncalabrese
Mar 30 2016 05:45
mostly pump, except for the 4g
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 05:47
strange
you worried about a low?
or expecting one i should say.
Jason Calabrese
@jasoncalabrese
Mar 30 2016 05:48
he had been running around thought the snack might take too long
but if we would have calibrated it would have been very different
Ben West
@bewest
Mar 30 2016 06:50
you south pawed, @CrushingT1D?
my watch goes on my left :-D
Jason Calabrese
@jasoncalabrese
Mar 30 2016 06:58
@bewest trying to get the share ble going again on an edison, docs say to edit init.d/bluetooth, /etc/init.d/bluetooth? Seems like I need to install something else
$ sudo apt-get install bluetooth
Reading package lists... Done
Building dependency tree
Reading state information... Done
The following extra packages will be installed:
  bluez
Suggested packages:
  bluez-cups bluez-obexd
The following NEW packages will be installed:
  bluetooth bluez
0 upgraded, 2 newly installed, 0 to remove and 57 not upgraded.
Need to get 787 kB of archives.
After this operation, 3,187 kB of additional disk space will be used.
Do you want to continue?
don't think I want a different version of bluez, but seems like I'm missing the the bluetooth service
Ben West
@bewest
Mar 30 2016 07:40
hmmm
is this systemd or upstart?
ah, edison is basicaly wheezy so upstart
what version fo bluez is that?
80% sure the distro supplied one is no good
benhamou
@benhamou
Mar 30 2016 08:54
hello I have troubles to acces my raspberry after I tried to make it paswordless
disconnected: no supported authentication methods available (server sent: publickey)
May one help me. I think have to alter something on winscp or putty ?
does anyone sell me a paradigm veo 722 <=Ver 2.6 pls
does anyone has experience with it?
for looping
tazitoo
@tazitoo
Mar 30 2016 12:07
@benhamou - I don't think there is a way for openaps to get info from, or send commands to the t:slim pump (regardless of G4 being present or not). I have a t:slim pump - changed to a medtronic for that reason
benhamou
@benhamou
Mar 30 2016 12:27
wiz https://github.com/mgranberry/YADLA it is possible to read but not to comand
Jason Calabrese
@jasoncalabrese
Mar 30 2016 13:00
Running 2 rigs last night worked great, no gaps or alarms
Matthias
@ceben80
Mar 30 2016 14:08
Today in the office....it is very impressive the APS :tada:
blob
John Sjolund
@sjolundjohn
Mar 30 2016 15:13

Currently stuck on 3.2.2.2.5. Check that you can communicate with your pump

All steps before that have been successful.

a) I assume Carelink USB needs to be attached to Raspberry to work.
b) I have not done any configuration of Carelink USB that I am aware of
c) I have personally not used this pump in my hand with Carelink or the USB stick (IE new pump for me)

Have I missed a step?

Thanks!

Scott Leibrand
@scottleibrand
Mar 30 2016 15:15
What error are you getting?
Dana Lewis
@danamlewis
Mar 30 2016 15:15
Have you installed deco care? And have you added a pump as a device?
Scott Leibrand
@scottleibrand
Mar 30 2016 15:16
@ceben80 and @jasoncalabrese: nice! :-)
John Sjolund
@sjolundjohn
Mar 30 2016 15:16

@scottleibrand not getting any error, not result at all
@danamlewis I do not believe i have installed Deco Care. I have installed pump as a device in step 3.2.2.2.2 which was successful.

So.. install deco care I guess?

Scott Leibrand
@scottleibrand
Mar 30 2016 15:17
decocare comes from quick-packages.sh
Dana Lewis
@danamlewis
Mar 30 2016 15:18
You are referring to numbers which makes me think you are looking at at branch version of the docs. I would make sure you are looking at dev or Dana-Dev for now.
John Sjolund
@sjolundjohn
Mar 30 2016 15:18
Ok - that was installed when I setup openaps and dependencies I suppose. Needs customization?

@danamlewis I followed the read the docs standard instructions- and associated files from GIT.

Question: As I embark on the journey, but likely weeks away before I would consider if I could actually use it, should I use the standard OR dev OpenAPS?

Dana Lewis
@danamlewis
Mar 30 2016 15:22
What command are you running, and what error are you getting?
I have a PR open to pull some updated docs into dev. After that comes through, I recommend using dev. (That's the Dana-dev branch now that's being PRed)
John Sjolund
@sjolundjohn
Mar 30 2016 15:27
pi@raspberrypi:~/my_openaps $ open use XXX model
Couldn't get a file descriptor referring to the console
Scott Leibrand
@scottleibrand
Mar 30 2016 15:27
As far as the master vs. dev branch of oref0/openaps, all versions of the docs recommend starting with master (via quick-packages.sh) and only moving to dev (quick-src) after you have a working overnight loop and are ready to try more advanced features.
Have you confirmed that the USB cable works? lsusb should show a Medtronic device.
John Sjolund
@sjolundjohn
Mar 30 2016 15:31
@scottleibrand I put Carelink straight in
*Carelink USB straight into the Raspberry
I will try command and report back - time to head home and put the twins to sleep. Thanks so far!
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 15:35
@bewest i am right handed. my watch goes on my left, but I normally eat with my right hand.
Jason Calabrese
@jasoncalabrese
Mar 30 2016 15:37
you need a watch for each arm, I started doing that awhile ago when I thought I was going to switch to Android Wear from Pebble, but now like both
the android watch is like a remote for my phone, and the pebble is great for glanceable data
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 15:38
yeah. another thing that sounds rediculous now but then i will be doing at some point in the future :)
Jason Calabrese
@jasoncalabrese
Mar 30 2016 15:40
then if the sensors notice you're eating with 2 hands you know it's a big sandwich
Scott Leibrand
@scottleibrand
Mar 30 2016 15:42
:-p
Matthias
@ceben80
Mar 30 2016 15:42
@jasoncalabrese ........ :+1: great :smile:
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 15:44
:laughing:
Dana Lewis
@danamlewis
Mar 30 2016 15:45
By the way all if you missed it yesterday - big PR for your review: openaps/docs#114
Need some reviews, any feedback, and thumbs up if it looks good to you
(Feedback in the PR thread would be great so we have a record of any notes tied to that PR)
Matthias
@ceben80
Mar 30 2016 15:49
btw - Wearing the APS (pi with a huge powerbank) in the waist belt is currently to be viewed critically so I wear it under the sweater. No one knows that it is a medical equipment.
Can really be misinterpreted :astonished:
Dana Lewis
@danamlewis
Mar 30 2016 15:51
@channemann has a camera case he fits his in that looks normal-ish
Chris Hannemann
@channemann
Mar 30 2016 16:17
@danamlewis normal-ish? That thing is sexy as hell!
Scott Leibrand
@scottleibrand
Mar 30 2016 16:23
:-p
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 16:34
:handbag:
Matthias
@ceben80
Mar 30 2016 16:37
I have a bag with jeans look. The makes me just 20 years older non 30.......lol
blob
:astonished:
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 16:45
:purse:
I saw a guy with a fanny pack yesterday and my first reaction was…wonder if he is running OpenAPS. Why else would he have that? haha
Matthias
@ceben80
Mar 30 2016 16:46
:smile:
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 16:51
does AMA adjustments appear anywhere in the updates sent to the OpenAPS pill in NS?
also for Auto Sens
Jason Calabrese
@jasoncalabrese
Mar 30 2016 16:53
you'll see the cob in the pill, but not the autosens yet
what does auto sens calc for your ratio, I haven't seen anything other than 1 in 3 days
and I don't see carbs decay
Scott Leibrand
@scottleibrand
Mar 30 2016 17:04
Something is not right with your setup, but not sure what.
Dana Lewis
@danamlewis
Mar 30 2016 17:04
@channemann it doesn't say that in the docs ;)
Scott Leibrand
@scottleibrand
Mar 30 2016 17:05
@jasoncalabrese need to drill down and debug why no carb decay
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 17:06
i fipped it on last night and was just going to start looking to see if it had done anything. i see it being more aggressive in high and low temping, but was not sure about auto sens. need to look in the logs i guess.
Jason Calabrese
@jasoncalabrese
Mar 30 2016 17:13
I'm really curious to see it looking right for someone else, doesn't feel like a config issue to me
I'm expecting a sort bug or something like that
It fits perfectly for the HotSwap!
Scott Leibrand
@scottleibrand
Mar 30 2016 17:24
:+1:
Dana Lewis
@danamlewis
Mar 30 2016 17:24
Niiiice.
@amazaheri and @sulka and people with 3D printers ^^
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 17:25
:shipit:
Matthias
@ceben80
Mar 30 2016 17:29
Cool @julza78
Chris Hannemann
@channemann
Mar 30 2016 17:33
@danamlewis should I target dev with that PR, or should I start a new /wip/sexy-as-hell branch?
Dana Lewis
@danamlewis
Mar 30 2016 17:34
Lol. I just went on a branch deleting spin yesterday (and will delete Dana-dev once people review 114 hint hint ;)) so build a branch at your own risk ;)
Chris Hannemann
@channemann
Mar 30 2016 17:41
+1
Jason Calabrese
@jasoncalabrese
Mar 30 2016 17:45
what's the best way to review that, the diff is huge
Dana Lewis
@danamlewis
Mar 30 2016 17:49
It's a big PR. Not sure of a better way other than scrolling all the way through to compare?
Or just click around the Dana-dev branch to see what the re-org looks like?
Scott Leibrand
@scottleibrand
Mar 30 2016 17:50
maybe post the RTD link to dana-dev in the comments?
Dana Lewis
@danamlewis
Mar 30 2016 17:56
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 18:07
Ah, my log is reporting "Optional feature Meal Assist not configured."
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 18:14
looks like i have everything setup, but must have a problem somewhere
Scott Leibrand
@scottleibrand
Mar 30 2016 18:16
probably a mismatch in report names
what does meal.json look like?
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 18:22
it’s empty right now.
thought it had something last night.
i will double check all the report names
yeah, issue i missed..
I am getting basalprofile_data is not defined
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 18:27
also, with tracking uptime of loop is there any way to see on a zero temp if it is running? there is no area to see light vs darker blue and i assume it can send it once and go down for 25 minutes
Peter Miller
@ochenmiller
Mar 30 2016 18:28
Another great visit with the endo! Thank you openaps and nightscout communities!
Scott Leibrand
@scottleibrand
Mar 30 2016 18:29
looking at light vs. dark only tells you if the loop has been running within the last 30m, so it's the same if a zero temp is set
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 18:29
whatever part of the AMA branch that is running is doing a great job. feels like it is making more between adjustments earlier than it was with dev
oh
Scott Leibrand
@scottleibrand
Mar 30 2016 18:29
one big thing I changed was forecasting 15m deviations for 30m into the future instead of just 15m
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 18:29
it’s way better :+1:
i dont smile much, but when I do….OpenAPS is crushing it.
Happy T1D.png
Dana Lewis
@danamlewis
Mar 30 2016 18:32
:)
Scott Leibrand
@scottleibrand
Mar 30 2016 18:42
:-D
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 18:58
So i think i have it working now, but I am seeing "reason": "COB: 34, Dev: -2, BGI: -0.7, Target: 95; Eventual BG 90>90 but Avg. Delta -2.00 < Exp. Delta -0.6; setting current basal of 2 as temp”,
i dont have 34 carbs on board. Might be same as what @jasoncalabrese is seeing?
Dana Lewis
@danamlewis
Mar 30 2016 18:59
Hm. Any carbs in recent activity?
(Trying to figure out if stale 34 or miscalculated 34 )
Jason Calabrese
@jasoncalabrese
Mar 30 2016 19:00
what I see is all carbs for DIA*1.5 counting as cob with no decay
@danamlewis you see it decay?
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:00
I put 17 carbs in my pump 2:45 mins ago
that’s it
Dana Lewis
@danamlewis
Mar 30 2016 19:01
17x2 is 34 so that's interesting
2h45 or 2min45 seconds?
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:02
hours
This message was deleted
This message was deleted
carb json
[{"_id":"56fbff3918301baf7c27a39b","bolus":{"_type":"Bolus","_description":"Bolus 2016-03-30T09:10:09 head[4], body[0] op[0x01]","timestamp":"2016-03-30T09:10:09-07:00","_body":"","appended":[{"_type":"UnabsorbedInsulinBolus","_description":"UnabsorbedInsulinBolus unknown head[5], body[0] op[0x5c]","_body":"","_head":"5c0528bcc0","data":[{"amount":1,"age":188}],"_date":""}],"programmed":1.8,"duration":0,"amount":1.8,"_head":"01121200","type":"normal","_date":"09ca291e10"},"timestamp":"2016-03-30T09:10:09-07:00","created_at":"2016-03-30T09:10:09-07:00","carbs":17,"ratio":"9","wizard":{"unknown_byte[8]":0,"_type":"BolusWizard","bg":0,"_byte[5]":0,"unknown_byte[10]":0,"_description":"BolusWizard 2016-03-30T09:10:09 head[2], body[13] op[0x5b]","timestamp":"2016-03-30T09:10:09-07:00","_body":"115009195a001200000000125a","bg_target_high":90,"sensitivity":25,"carb_ratio":9,"food_estimate":1.8,"unabsorbed_insulin_total":0,"correction_estimate":0,"carb_input":17,"_head":"5b00","unabsorbed_insulin_count":"??","_byte[7]":0,"bolus_estimate":1.8,"_date":"09ca091e10","bg_target_low":90},"eventType":"Meal Bolus","insulin":1.8,"notes":"Normal bolus with wizard.\nProgrammed bolus 1.8\nDelivered bolus 1.8\nPercent delivered:  100%\nFood estimate 1.8\nCorrection estimate 0\nBolus estimate 1.8\nTarget low 90\nTarget high 90\nHypothetical glucose delta -45","medtronic":"mm://openaps/mm-format-ns-treatments/Meal Bolus","enteredBy":"openaps://medtronic/722"}]
the interesting part might be “unabsorbed_insulin_count: “??"
Dana Lewis
@danamlewis
Mar 30 2016 19:04
Huh, yea; and up top where it notes appended with unknown head
I'm across the street with out my loop and apparently hotel wifi bonked on the rigs there so can't compare my logs at the moment
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:06
i think that file is pulled from NS though
i am not sure how the algo takes that into accout vs the history file
Dana Lewis
@danamlewis
Mar 30 2016 19:07
Do you normally enter carbs via pump, or NS?
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:07
pump
Dana Lewis
@danamlewis
Mar 30 2016 19:07
Wondering if there's something for how we get carbs from DIYPS to NS to algo
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:07
i wonder if it is counting both the pump carbs and NS carbs. hence the 2x.
Dana Lewis
@danamlewis
Mar 30 2016 19:07
@jasoncalabrese are you still entering carbs on pump or just NS?
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:07
but still not sure why its not decaying either
Dana Lewis
@danamlewis
Mar 30 2016 19:07
It should be dedup
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:09
will be interesting what happen goes past 3 hours in a few mins
Dana Lewis
@danamlewis
Mar 30 2016 19:10
We made a change because of seeing carbs kick in at hour 5 after a meal
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:11
yeah, but that report in the ini only pulls the last 3 hours of time stamps
so it’s going to lose it next pull from NS
Dana Lewis
@danamlewis
Mar 30 2016 19:12
Ah
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:16
ok so my carb history file is blank now
its still saying 34 COB, but will give it 1-2 more loops to confirm
Scott Leibrand
@scottleibrand
Mar 30 2016 19:19
in the mean time, I would run oref0-meal manually and see if you can repro. maybe uncomment some of the debug console.error statements to see what it's doing...
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:19
still thinks 34, so it must be using the pump number and not the NS (aka deduping it) so that is good
Jason Calabrese
@jasoncalabrese
Mar 30 2016 19:22
Only on the pump today, I'm not pulling carbs from NS yet
Dana Lewis
@danamlewis
Mar 30 2016 19:24
Ok. So both coming from pump entry then
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:31
have not tried to uncomment anything yet. but first run i noticed if i pass in the pump-history.json it showed 17 carbs, and if i pass in pump-history-zoned.json it shows 0. (note, i am using the zoned in my ini, but this could be a possible problem if people set it up wrong)
i also did not wait long enough, my log now shows 17 carbs instead of 34.
it dropped about 3hrs 5 minuts after carb entry on the pump. and at 3hrs the pull from NS dropped off.
i did not double enter it. that is just treatments that NS knew about
Scott Leibrand
@scottleibrand
Mar 30 2016 19:34
how are you downloading treatments from NS?
we do openaps alias add ns-meal-carbs '! bash -c "curl -m 30 -s \"$NIGHTSCOUT_HOST/api/v1/treatments.json?find\[created_at\]\[\$gte\]=date -d \"6 hours ago\" -Iminutes&find\[carbs\]\[\$exists\]=true\" > monitor/carbhistory.json && oref0-meal monitor/pumphistory-zoned.json settings/profile.json monitor/clock-zoned.json monitor/carbhistory.json > monitor/meal.json; exit 0"'
the important part there is we only use the carbhistory if the pumphistory has zero carbs
may need to put dedupe logic into oref0-meal instead
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:36
yeah i am not doing that. i just followed the code i was given awhile back
which is
pull-ns-meal-carbs = "! bash -c \"curl -m 30 -s \\\"$NIGHTSCOUT_HOST/api/v1/treatments.json?find\\[created_at\\]\\[\\$gte\\]=`date -d \\\"3 hours ago\\\" -Iminutes`&find\\[carbs\\]\\[\\$exists\\]=true\\\" > monitor/carbhistory.json; exit 0\""
i actually thought the back end code figured it out. What are you using @jasoncalabrese
so i am only doing 3 hours
but when i run it i am passing in carbhistory and pump history (zoned).
so if i enter carbs in the pump and then do another snack and only do carb entry in NS but not the pump then it’s not going to work correctly?
(i dont normally do that, but i might)
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:42
Ben West
@bewest
Mar 30 2016 19:42
ns-get might be easier than a raw curl device
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 19:44
currently when i run oref0-meal by hand it is outputting 0 carbs, but my loop is including 17 carbs in it’s decision reason still
nothing new or interesting got reported when i uncommented out all the console logs
Jason Calabrese
@jasoncalabrese
Mar 30 2016 20:43
I'm not pulling carbs from NS at all yet, wanted to make sure it was right with pump history only first
Jim is CrushingT1D
@CrushingT1D
Mar 30 2016 21:11
even better.
Ben West
@bewest
Mar 30 2016 21:41
tweaked the ns tools a bit
should be a lot easier to use now I think

Something like this:

$ nightscout autoconfigure-device-crud myhost.com foboar
added process://ns/nightscout/ns NIGHTSCOUT_HOST API_SECRET
process://ns/nightscout/ns myhost.com 143505794bdd283f7daed50f42d201b6926892c8

Now allows this as a result:

openaps use ns shell get entries.json "count=10"

openaps use ns shell upload treatments.json recently/created-treatments.json
some more info here openaps/oref0#93
Scott Leibrand
@scottleibrand
Mar 30 2016 22:08
someone want to give that a try and update the docs to use that method?
Dana Lewis
@danamlewis
Mar 30 2016 22:08
^ :+1:
Target Dana-dev might be easiest if it's before 114 gets merged to dev
Ben West
@bewest
Mar 30 2016 22:16
well, it's in an experimental branch, so docs will need to track it via dev -> release
btw, thinking of doing another openaps release soon, lots of bugs and bug fixes in dev branch, as well as some api changes that probably shouldn't sit in dev too long
can do similar script to setup oref0 devices, might be good to have few more scripts to shape groups of reports with common names/profiles
Ben West
@bewest
Mar 30 2016 22:22
would like to start getting more rigorous about sending most device data to a foo-raw.json file as best practice, then running some tool on it always, eg units/timezone, etc
Jason Calabrese
@jasoncalabrese
Mar 30 2016 22:35
think I have to go back to dev while debugging the cob issue, seen a few lows that might be partially related, hard to say for sure
Scott Leibrand
@scottleibrand
Mar 30 2016 22:53
yeah, that's probably safest. I don't know how much extra insulin it might give if it thinks COB is higher than it really is. it won't be a ton, as it's still based on deviations, but maybe enough to contribute to a subsequent low.
Jason Calabrese
@jasoncalabrese
Mar 30 2016 22:55
overall the extra aggressiveness has been good, but a couple times I think it's gone the other way
the good thing is it's easy to switch back and forth
the extra fields for AMA don't get in the way so the ini doesn't have to change
I do really want it to work and then want to build on it
was thinking about sending negative carbs from NS for meals that weren't finished
Dana Lewis
@danamlewis
Mar 30 2016 22:58
why not just edit the actual carb amount? undo, and redo with the correct amount essentially?
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:10
we'd enter the planned amount in the bolus wizard, then maybe realize it's not needed or wanted
not sure the best way to handle how breakfast is stretched out waiting for good time for the milk
it's like a really long DIA in the morning
Dana Lewis
@danamlewis
Mar 30 2016 23:11
food time: do carbs for food and the full bolus you think; then when time to drink the milk, put in milk carbs so the decay and activity is correct.
Ben West
@bewest
Mar 30 2016 23:12
ah, this gets at the scheduling-insulin idea I had earlier, except it's scheduling carbs
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:13
need to get the bolus for the milk going way way before
Ben West
@bewest
Mar 30 2016 23:13
would be nice to be able to interact with openaps somehow on basis of the expectations "we expect a rise/fall of this shape and if/when that happens we're doing these carbs"
Dana Lewis
@danamlewis
Mar 30 2016 23:13
right. but there’s no reason to not just do the right amount of bolus and then enter carbs later as they’re ingested
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:14
either that or his morning carb ratio is a really crazy 6, but and later in the day is 18
Ben West
@bewest
Mar 30 2016 23:14
right inverse is schedule insulin in this way (and it's not a simple temp/bolus combo)
Dana Lewis
@danamlewis
Mar 30 2016 23:14
scheduling carbs is dangerous if something happens to delay and the user isn’t able to adjust the system to know that
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:14
it's a time bomb
Dana Lewis
@danamlewis
Mar 30 2016 23:18
30g breakfast + knowing 15g milk with a 1:10 ratio would be entering 30 carbs and up arrowing to a 4.5 bolus. then when the carbs are ingested, a 15g carb entry. (making up numbers, but that’s the methodology I would use)
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:19
but the button pusher at that time of the day just wants to enter all the carbs
I call them phantom carbs
Dana Lewis
@danamlewis
Mar 30 2016 23:19
so enter the milk carbs, but do zero bolus. what other carbs are you talking about?
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:19
but need a way to let the system know about it
Dana Lewis
@danamlewis
Mar 30 2016 23:20
which carbs are you talking about? can you give an example if the above one isn’t accurate for the situation?
Ben West
@bewest
Mar 30 2016 23:21
it's UX issue
you have operators attention once and only once
Dana Lewis
@danamlewis
Mar 30 2016 23:21
right. but I want to clarify the situation of WHO is the operator and WHEN and WHAT action is being taken by whom.
drink milk, enter carbs, do no bolus because bolus was performed with breakfast.
yes? or no?
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:22
breakfast is 24g, carb ratio then is 10, so bolus is 2.4
right now the carbs for the milk aren't getting entered at the time he has it, since they're already entered
Dana Lewis
@danamlewis
Mar 30 2016 23:22
how many carbs are milk?
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:23
Screenshot 2016-03-30 16.23.24.png
12g for milk and 12g for the rest of breakfast
all entered at once
he had the milk by the red dots around 9am
Dana Lewis
@danamlewis
Mar 30 2016 23:24
so…at breakfast time, enter 24 carbs, but do a 3.6 bolus (24+12 carbs of milk worth). at milk time, enter 12 carbs and do no more bolus. Unless another snack is being eaten and mixed into this?
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:25
24g is total with milk
Dana Lewis
@danamlewis
Mar 30 2016 23:26
so if breakfast food is 12…. enter 12 carb and do a 2.4 bolus (food+milk carbs), then only enter 12 carbs of milk later with no bolus. ?
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:26
part of the issue we're having is the bolus wizard is adding a correction does, when the loop probably already added that amount or more
Ben West
@bewest
Mar 30 2016 23:27
medtronic would tell you to try dual wave I guess
ah, you can dial the insulin to 0
Dana Lewis
@danamlewis
Mar 30 2016 23:27
then do 2.3 bolus upfront, and 12 carbs with a .1 at milk time?
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:27
we want all the insulin, but waiting on the carbs
yeah, could do that, but a pain
lots of button pushing
Dana Lewis
@danamlewis
Mar 30 2016 23:27
or do a 2.4 and then a .1 knowing the loop will counteract the .1 if needed
Ben West
@bewest
Mar 30 2016 23:27
yeah, UX
Dana Lewis
@danamlewis
Mar 30 2016 23:28
that’s a problem with the pump
Ben West
@bewest
Mar 30 2016 23:28
yeah, not saying it's our UX issue
Dana Lewis
@danamlewis
Mar 30 2016 23:28
can’t circumvent that if you want the loop to do the smartest thing, it needs to know about carbs one way or another :/
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:28
but if I enter -12g later
effect would be the same
and I could do it after the morning rush
Ben West
@bewest
Mar 30 2016 23:28
would be nice to communicate expecation that these are slow carbs
Dana Lewis
@danamlewis
Mar 30 2016 23:28
much higher risk of everything swiss cheese-ing, though
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:29
yeah, would be nice if he'd eat a different breakfast
no carbs = no problems
Ben West
@bewest
Mar 30 2016 23:29
don't get diabetes ;-)
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:30
or that
Dana Lewis
@danamlewis
Mar 30 2016 23:30
or pull in carbs from NS, and have nurse do the carb on NS and that way you don’t have to pull in the 0.1 bolus from the wizard
yea, that
or turn into a mouse
because mice’s diabetes is easily cured :/
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:31
there way too much coordination needed for mornings now
Ben West
@bewest
Mar 30 2016 23:31
yeah, and, this is a case where you have reasonable expectations/mental model
and there's just not tech to capture it
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:31
to add to that, he really wants the milk, so some even if it looks like he doesn't need it, I tell him to go get it and do an easy bouls
Ben West
@bewest
Mar 30 2016 23:32
some way to re-introduce dual wave support somehow might be nice
although, not sure that would do it here
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:32
that's opposite
Ben West
@bewest
Mar 30 2016 23:32
yeah
Dana Lewis
@danamlewis
Mar 30 2016 23:33
this might be one of the situations where you deal with more work, knowing it’s a pain and a workaround, or adjust expectations of BG behavior knowing the tech doesn’t support the optimall solution :/
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:34
or send the bolus (after we fix the triple enact), can't see how a smart bolus wizard that doesn't do the bolus would work
lots of these issues come from using the bolus wizard on the pump
Dana Lewis
@danamlewis
Mar 30 2016 23:34
yep
we built in undo carbs on diyps to handle this situation. but i don’t use the bolus wizard to enter my stuff, so it’s easy to delete and re-record the carbs regardless of what the insulin was
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:36
thought about entering only carbs, and having the system ask for a bolus
but clunky
there the really nice edit mode in NS where you can move carbs around
would need a split
"or adjust expectations of BG behavior knowing the tech doesn’t support the optimal" is what I'm hearing some someone else, but not something I'm going to accept
Scott Leibrand
@scottleibrand
Mar 30 2016 23:40
have you tried just entering all the carbs up front, for breakfast plus milk?
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:41
thats what we do now
Scott Leibrand
@scottleibrand
Mar 30 2016 23:41
seems like once we get your AMA carb decay fixed, that will work pretty well
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:41
but then have to watch to find the right time, send messages, etc and then the data is still wrong
Dana Lewis
@danamlewis
Mar 30 2016 23:42
@jasoncalabrese yep. everyone has to figure out what they’re willing to do. complicated by the fact that you have multiple people who support data entry into the system, vs a single primary user.
how does the split carb feature (or edit) work when there’s pump records also going in? or sounds like you might not have tested that?
Ben West
@bewest
Mar 30 2016 23:43
partly trying to figure out what to design next, @danamlewis ;-)
Dana Lewis
@danamlewis
Mar 30 2016 23:44
@bewest which piece(s) do you mean?
Ben West
@bewest
Mar 30 2016 23:44
whatever shakes out of exploring what's needed and what's possible, regardless of how unreasonable it may seem ;-)
but the thing I'm thinking about is using touch interface on a phone or maybe a watchface app to influence things in openaps/pump/NS
Dana Lewis
@danamlewis
Mar 30 2016 23:45
yea. we should talk through some of the plumbing and de-dupe we’ve done between DIYPS and pump uploading
because it sounds like some of the same is needed with NS and pump uploading
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:45
I think if we could merge carbs from NS and the pump in the right way it could work
then NS is where we can the carb amounts timing just right (if needed)
Ben West
@bewest
Mar 30 2016 23:46
yeah, but getting greedy, even putting data into NS is too much work for me
I like @loudnate's applewatch thingy
Dana Lewis
@danamlewis
Mar 30 2016 23:46
@bewest yes. i want to support people putting data in one place. for me, that’s diyps. for others, NS. for others, watch. for others, pump. etc.
but should be able to pick your poison and have it work
Ben West
@bewest
Mar 30 2016 23:47
yeah
Dana Lewis
@danamlewis
Mar 30 2016 23:48
it’s partially a matter of whether you can “delete” a pump record that’s combined insulin and carbs, and rewrite as just the insulin if you end up adjusting the carbs
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:48
thats where I was thinking -carbs
it would be an adjustment on top of the pump history
Dana Lewis
@danamlewis
Mar 30 2016 23:49
which we haven’t touched because i’m not using the bolus wizard as primary, so we don’t have to muck with actual pump records, just manual carb entry from (choose your place)
but it’s adding increased error opps if you have to delete and re-write a record, because of also watching for double entry, etc. and no longer having “one source of truth” of what happened
so, that makes me nervous
Jason Calabrese
@jasoncalabrese
Mar 30 2016 23:50
maybe they should be treated as different things
Dana Lewis
@danamlewis
Mar 30 2016 23:50
they can be if they’re entered from (choose your place), but if it’s pulled from the pump it’s tied
Nathan Racklyeft
@loudnate
Mar 30 2016 23:53
I've found myself revising carbs a few times in Naterade, but HealthKit remains my source of truth for all data going back to midnight of the previous day. Data older than that goes to a server.
Dana Lewis
@danamlewis
Mar 30 2016 23:54
what’s your process for revising, @loudnate ?
Nathan Racklyeft
@loudnate
Mar 30 2016 23:55
HealthKit records are immutable, so it's a delete and recreate. That's transparent to the user modifying the entry on the phone though.
The reasons for revising are usually either "gosh one bite in and I don't want to eat the rest" or "my wild guess on that restaurant plate was too low/too high"
Dana Lewis
@danamlewis
Mar 30 2016 23:57
but you’re not doing any bolus wizard pump entry, correct?
Ben West
@bewest
Mar 30 2016 23:58
he was trying to :-D we think it's possible but don't think we've mastered the technique yet
Nathan Racklyeft
@loudnate
Mar 30 2016 23:58
No, my pump buttons are only used for prime/rewind nowadays
Dana Lewis
@danamlewis
Mar 30 2016 23:58
:) @loudnate
Ben West
@bewest
Mar 30 2016 23:58
well, not through the buttons ;-)
Nathan Racklyeft
@loudnate
Mar 30 2016 23:58
Yeah I nearly bricked one looking for JournalEntryMealMarker
Well bricked /= accidentally learned the screen contrast command
Dana Lewis
@danamlewis
Mar 30 2016 23:59
haha
Ben West
@bewest
Mar 30 2016 23:59
yeah, that's actually one of the more devastating commands, it turns out
Scott Leibrand
@scottleibrand
Mar 30 2016 23:59
I wonder why they have contrast go all the way down to zero
Nathan Racklyeft
@loudnate
Mar 30 2016 23:59
I think I put it into decocare? If not I should.