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

15th
May 2016
Paul Martin
@Spazholio
May 15 2016 00:22
Trying to loop at a water park. Can’t really recommend it. Yeesh.
Dana Lewis
@danamlewis
May 15 2016 00:24
I always find after my sensor is submerged for a while that it reads faslely high for up to an hour. So in between swims doesn't help much for looping, but does when you're finished.
Paul Martin
@Spazholio
May 15 2016 00:25
I’ve been bottoming like crazy. No pump attached, yet I’m bottoming. Go figure.
And this Keto Chow I’m trying, while delicious, might be causing issues despite the claim of 5-10g of digestible carbs.
But it’s tasty, so it bears more investigation. =)
Dana Lewis
@danamlewis
May 15 2016 00:26
Well, that's activity for you!
Keep in mind you might want to set a several hour temp to zero and then suspend, so you pump will have accurate idea of net iob when you reattach later
Paul Martin
@Spazholio
May 15 2016 00:28
Well, I never usually swim, but I got a deal for a water park and my kids love it, so here we are. I just disconnected everything, but went back to the locker roughly every hour to check levels and all that. The pump was suspended, so should the net IOB be fairly accurate? I’ve been reattched for a bit now, and I think things are ok.
Dana Lewis
@danamlewis
May 15 2016 00:30
Only if you set a temp to zero, otherwise it doesn't know you were off it for an hour. Just means your true net is a smudge lower than it is calculating now, so if you end up with an unexpected rise in the next few hours, that's usually why but shouldn't be a big deal with a loop to catch the rise.
Paul Martin
@Spazholio
May 15 2016 00:36
Ah, ok. Interesting - so if it detects the pump is suspended, does it not take that into account, or does it take it into account in a different way?
Dana Lewis
@danamlewis
May 15 2016 00:41
It does not take it into account. I think Nate has a way to read it and calculate it, but no one has integrated it into oref0. So for now, temp to zero and suspend. (That's what I do for showers, too)
Paul Martin
@Spazholio
May 15 2016 00:41
Ahh, ok. Thanks for the heads up.
Dana Lewis
@danamlewis
May 15 2016 00:42
:+1:
Scott Leibrand
@scottleibrand
May 15 2016 00:44
Nate's mmhistorytools do that and a few other things we really should be doing. If someone wants a project, integrating mmhistorytools with oref0 would be a good one.
Paul Martin
@Spazholio
May 15 2016 00:49
I’ll do it! Just a few things first - what is mmhistorytools? =)
(sorry, I think the chlorine’s affecting me)
Scott Leibrand
@scottleibrand
May 15 2016 01:00
All: still looking for a bit more feedback on releasing oref0 dev to master: openaps/oref0#105 - if you’ve been running dev, or have time to switch over to it to sanity check everything, I would appreciate a few more :+1: ’s before we merge it and do a release, hopefully tomorrow.
Scott Leibrand
@scottleibrand
May 15 2016 01:43
@Spazholio also openaps/oref0#66
Eric Luhrs
@eluhrs
May 15 2016 02:14
👍🏼 Have been running dev for at least a few weeks with no problems. Currently only using auto sensitivity, not advanced meal assist.
Added thumbs up to PR too.
Scott Leibrand
@scottleibrand
May 15 2016 02:16
Thx. If we can get a few more we might even be able to do a release tonight. :-)
ddaniels1
@ddaniels1
May 15 2016 03:03
Screenshot 2016-05-14 20.02.42.png
Scott Leibrand
@scottleibrand
May 15 2016 03:05
What is your question/comment about that?
ddaniels1
@ddaniels1
May 15 2016 03:09
What does "undefined boluses and undefined target" mean?
Also,
Scott Leibrand
@scottleibrand
May 15 2016 03:09
that's Carbs: undefined and Boluses: undefined. means you don't have meal assist set up
ddaniels1
@ddaniels1
May 15 2016 03:10
I think I have autosensitivty working and working on a PR for the docs for that to clarify the configurations. Is there a way to tell it is operating on the NS interface?
Scott Leibrand
@scottleibrand
May 15 2016 03:11
I just added something to AMA that displays ISF, but there's nothing in dev that displays autosens info in NS
you'll need to look at the contents of your json or log files on the pi itself
or I suppose I could cherry-pick that commit into dev...
ddaniels1
@ddaniels1
May 15 2016 03:14
pi@raspberrypi:~/Documents/try5 $ openaps report invoke enact/suggested.json determine-basal://text/shell/enact/suggested.json {"delta":1,"glucose":128,"avgdelta":1.333} {"duration":3,"rate":1.35,"temp":"absolute"} {"netbasalinsulin":0.05,"activity":-0.0009,"basaliob":0.164,"hightempinsulin":0.45,"bolussnooze":0,"iob":0.171} {"max_iob":2,"type":"current","dia":3,"current_basal":0.45,"max_daily_basal":0.45,"max_basal":1.55,"min_bg":110,"max_bg":130,"sens":89}
Here is my suggest.json, doe sit look right?
Scott Leibrand
@scottleibrand
May 15 2016 03:14
what is your normal ISF?
and what does autosens.json or similar show for the ratio?
ddaniels1
@ddaniels1
May 15 2016 03:15
set now at 89
cd ..
oops
ddaniels1
@ddaniels1
May 15 2016 03:22
Here is my auto-sens.json
p=0.60: 1.00, -2.40, 6.38
p=0.58: 1.00, -2.98, 6.03
p=0.56: 0.67, -3.60, 5.32
p=0.54: 0.67, -4.06, 4.61
p=0.52: 0.33, -4.61, 3.49
p=0.50: 0.00, -5.07, 3.17
p=0.48: 0.00, -5.36, 2.75
p=0.46: -0.33, -5.70, 2.05
p=0.44: -0.67, -6.05, 1.70
p=0.42: -1.00, -6.42, 1.18
p=0.40: -1.33, -6.76, 0.98
p=0.38: -2.00, -7.11, 0.55
p=0.36: -2.33, -7.43, 0.28
p=0.34: -2.67, -7.79, 0.01
p=0.32: -3.00, -8.13, -0.34
p=0.30: -3.33, -8.43, -0.49
p=0.28: -3.40, -8.78, -0.90
p=0.26: -3.67, -9.07, -1.18
Mean deviation: 4.76
Sensitivity within normal ranges
Basal adjustment 0.00U/hr
Ratio: 100%: new ISF: 89.0mg/dL/U
reporting monitor/auto-sens.json
Scott Leibrand
@scottleibrand
May 15 2016 03:23
that looks reasonable
ddaniels1
@ddaniels1
May 15 2016 03:36
I read #99, can you explain again what the 4 columns represent? and how is autosens different from the loop reacting to the BGI varying from expected?
Scott Leibrand
@scottleibrand
May 15 2016 03:44
#99 is a new thing we haven't done yet. the algorithm you're seeing is openaps/oref0#58
you can ignore the middle two columns. the interesting thing is what percentile (first column) corresponds to zero. in your case, that's p=0.34.
in AMA the output is a lot easier to read, and the algorithm is a lot better at actually detecting sensitivity or resistance, because it excludes big deviations (usually caused by carb absorption) and shoots for 45-50% of the deviations being negative, instead of 30-50% like in dev.
ddaniels1
@ddaniels1
May 15 2016 03:53
planning configuring meal assist next tonight, found this: https://gist.github.com/scottleibrand/63a8541397378bf9f4b8
Should I do this or AMA? and is there any help on that?
Scott Leibrand
@scottleibrand
May 15 2016 03:56
I would start with that first, as it's mostly also useful for AMA. Also, AMA is not as clean of an upgrade: switching to the advanced-meal-assist branch will break your loop until you configure all the new stuff, so the less new stuff you have to configure when you do so the better.
at some point after we get dev released to master, one of the things we need to work on is improving the upgrade process from dev to AMA.
Matthias
@ceben80
May 15 2016 04:21
Yeah.....a new Looper @C-Ville has arrived :star2: The openAPS community is growing in Germany :tada:
Dana Lewis
@danamlewis
May 15 2016 04:30
:confetti_ball: :clap:
ddaniels1
@ddaniels1
May 15 2016 04:34
thanks @scottleibrand, I figured out based on the master docs the gather alias that is suggested won't work with auto-sens because need monitor-cgm and gather-profile first as they contain elements needed to run the auto-sens repot, which I decided to put in the monitor-pump alias. Is this worth submitting as a PR or is this a individual configuration issue that people need to figure out on their own? If find a number of times if I deviate from something it causes a number of issues that I can sort out with time, and actually is a learning experience on how the whole thing works, but time nonetheless.
Scott Leibrand
@scottleibrand
May 15 2016 04:37
Are you saying the gist doesn't work, or what is in the main docs?
that gist is out of date, and needs to be replaced with a page in the main docs with steps that actually work and are readable on plain English.
Someone did that with autosens already I think: we need someone to do so for meal assist too.
ddaniels1
@ddaniels1
May 15 2016 04:56
sorry what is gist?
Scott Leibrand
@scottleibrand
May 15 2016 05:05
That URL you posted.
ddaniels1
@ddaniels1
May 15 2016 05:06
What is in the main docs.
Scott Leibrand
@scottleibrand
May 15 2016 05:08
You might just be running into a problem where autosens only starts working the second time you run it, because you need to get the settings and profile stuff once, which happens at the end of the loop.
ddaniels1
@ddaniels1
May 15 2016 05:10
Don't think so tried it a few times. My original gather alias was: `gather ! bash -c "rm -f monitor/*; openaps gather-profile && openaps monitor-cgm && openaps monitor-pump"
All data was removed from monitor/ to prevent use of old data
I put auto-sens in gather-profile and hence it didn't work. When I moved it to the end of the gather alias and the other data is there it all works
Works for me now, trying to find in the docs where the gather profile alias code is to double check that isn't just something i did on my own.
ddaniels1
@ddaniels1
May 15 2016 05:28
referring to your response @scottleibrand regarding the percentile that corresponds to 0 being 34% You mention in openaps/oref0#58 that 30-50% should be negative so that break point means the basal and or ISF is well tuned?
Dana Lewis
@danamlewis
May 15 2016 05:35
It means isf observed matches what's expected
ddaniels1
@ddaniels1
May 15 2016 05:41
ok thanks. I am submitting a PR tonight for the autosens configuration page in the as there were a number of things that didn't work as is. I am trying to be sure that it all flows from the loop that should have been setup. @danamlewis I have it forked from dev is that ok?
Dana Lewis
@danamlewis
May 15 2016 05:44
Yep, that's great - thanks for the pR!
Hristina Dimova
@xpucuto
May 15 2016 13:31
@danamlewis @Spazholio I think i have the latest dev and my Openaps automatically set the temp basal to 0U when I suspended the pump today ;)
Paul Martin
@Spazholio
May 15 2016 13:32
Alas, I didn’t pay too much attention, and uploading to NS is sporadic at best while I’m out and about.
Scott Leibrand
@scottleibrand
May 15 2016 15:04
@xpucuto OpenAPS can't set a temp after you suspend: the pump won't accept it.
William Fallon
@inform880
May 15 2016 16:51
Hi, I'm just thinking about possibilities in the future, and I'm slowly learning more about linux systems, but could you run openaps on a full blown debian machine?
Scott Leibrand
@scottleibrand
May 15 2016 16:52
yes, openaps will run on any linux or OS X machine. Carelink compatibility is limited, but TI stick and mmeowlink work well with everything.
William Fallon
@inform880
May 15 2016 16:53
OK, thanks.
William Fallon
@inform880
May 15 2016 17:02
What are the problems with running openaps on a android tablet or phone? You have wifi, bluetooth for the share, and potentially a mobile connection.
Scott Leibrand
@scottleibrand
May 15 2016 17:16
HAPP runs oref0 on Android. The pump driver to talk to the MDT pump via RileyLink isn't published yet though.
If we had pumps that did native Bluetooth we'd probably have moved to mobile by now. But they don't speak 900MHz, so you need something like a RL.
Also, LoopKit for iOS and @loudnate's naterade provide a decent framework for looping on iPhones.
But no one has yet integrated oref0 with that, and I haven't taken the plunge to learn swift yet.
And there is of course the question of whether it's better to have your phone or a dedicated device running your loop. As Ben puts it, he doesn't want Angry Birds killing his APS.
Scott Leibrand
@scottleibrand
May 15 2016 17:23
If we want to make OpenAPS accessible to hundreds of people rather than dozens, we may want to build an app, though. But lots of risks to address if we make it "too easy".
Paul Martin
@Spazholio
May 15 2016 17:25
When using a TI stick, is there some way to simulate unplugging and re-plugging the stick in without actually having to physically do it? I’m easily going hours without my preflight succeeding, and I’d rather not have to set a reminder to reseat the device every hour.
Scott Leibrand
@scottleibrand
May 15 2016 17:25
How is it connected (USB or serial)? And Pi or Edison?
Paul Martin
@Spazholio
May 15 2016 17:26
USB on Pi.
Scott Leibrand
@scottleibrand
May 15 2016 17:26
For USB from a Pi, you'll want to use oref0-reset-usb to power cycle the USB bus.
Paul Martin
@Spazholio
May 15 2016 17:27
Thanks. Ran it. Let’s see how the next loop goes.
Scott Leibrand
@scottleibrand
May 15 2016 17:27
something like * * * * * killall -g --older-than 7m openaps && sudo oref0-reset-usb in cron will do the trick
Paul Martin
@Spazholio
May 15 2016 17:27
Would it be appropriate to run that if the preflight ever fails?
Scott Leibrand
@scottleibrand
May 15 2016 17:28
that will run it if the loop gets stuck waiting for the stick to respond
not sure if that is still the problem with newer mmeowlink's though
Paul Martin
@Spazholio
May 15 2016 17:28
I mean, if the preflight fails, may as well restart the USB bus. Won’t disrupt anything since I usually don’t have the Dexcom physicall plugged in. Maybe that in addition to the cron, to catch the possible hung loops...
Scott Leibrand
@scottleibrand
May 15 2016 17:28
they might error out, in which case you'd need to differentiate errors due to stick vs. pump being out of range
I don't think you want to be resetting USB every 5 minutes when you're away from the Pi
Paul Martin
@Spazholio
May 15 2016 17:29
I want this loop to run every single time, but I acknowledge that’s unrealistic with all the variables, so I’m trying to to aggressively eliminate as many variables as I can without being TOO aggressive. Hence me asking for advice here. =)
Scott Leibrand
@scottleibrand
May 15 2016 17:29
more likely to get the Pi completely hung and require a full power cycle if you reset it that often
Paul Martin
@Spazholio
May 15 2016 17:30
Ouch, really? Ok, mental note to self...
Scott Leibrand
@scottleibrand
May 15 2016 17:30
for carelink, we normally do mm-stick warmup 2>&1 || sudo oref0-reset-usb
but I'm not sure if there is a similar "check the TI stick" tool yet
Paul Martin
@Spazholio
May 15 2016 17:31
Oskar would be the guy to bug about that?
Scott Leibrand
@scottleibrand
May 15 2016 17:31
could maybe grep the output of the preflight command and make sure it has the subg lines in it...
either @oskarpearson or @bewest would know if something like that already exists for mmeowlink
Paul Martin
@Spazholio
May 15 2016 17:32
Cool.
Scott Leibrand
@scottleibrand
May 15 2016 17:32
but if not, it might not require a new tool
probably the first thing to do is to characterize the broken behavior (how to detect it)
Paul Martin
@Spazholio
May 15 2016 17:33
I need to bug @bewest about uploading missing treatments (not entries) to NS. Hoping it’s really similar to what we did for filling in gaps in entries. I’m sure he’s overjoyed to see this, too. =)
Scott Leibrand
@scottleibrand
May 15 2016 17:33
read back through what he was telling you to do when he thought that's what you were trying to do. :)
now that you've gotten entries gap sync working, it'll probably make a lot more sense and be easier to duplicate for treatments
Paul Martin
@Spazholio
May 15 2016 17:34
Yep. But entries from from NS + Dexcom, right? If I’m understanding correctly, treatments are things I enter into the pump that didn’t make it to NS for whatever reasons - is that correct?
Scott Leibrand
@scottleibrand
May 15 2016 17:34
yeah, particularly the contents of pumphistory.json
Paul Martin
@Spazholio
May 15 2016 17:36
Aaaand, I’ve already got a pumphistory-24h-zoned.json file (that’s living in settings instead of monitor for some reason) so I likely already have the data I need to work this out...
Scott Leibrand
@scottleibrand
May 15 2016 17:38
that one is in settings so it doesn't get blown away each loop, since the data is still relevant for autosens and AMA COB calculation purposes for more like an hour
Paul Martin
@Spazholio
May 15 2016 17:39
Oh god, that’s on purpose? I thought I did something boneheaded again. whew =)
Scott Leibrand
@scottleibrand
May 15 2016 17:39
yeah, my stuff put it there. :)
William Fallon
@inform880
May 15 2016 17:45
@scottleibrand What do you mean by HAPP? Also, what about running a linux distro on the phone, and using a microusb to usb to plug in with a the TI stick?
Paul Martin
@Spazholio
May 15 2016 17:45
You know, I removed the “pushover” plugin from NS a while back, but it’s still sending me stuff. Do I need to yank the authorization from Pushover to make that stop until I have more finely tuned?
Lori
@LoriLori
May 15 2016 17:47
@inform880 phone running some linux distro would run 2 hours on battery
there is quite a lot of work to make it run 3 days which is not part of linux
William Fallon
@inform880
May 15 2016 17:49
@LorelaiL Alright
Lori
@LoriLori
May 15 2016 17:49
@inform880 HAPP is https://gitter.im/timomer/HAPP
and here is part of responsible for running oref0 on android https://github.com/LorelaiL/AndroidAPS it is integrated in happ
Stephan
@MosiGitHub
May 15 2016 18:06
@scottleibrand Hi, back from work today. So I did the PR regarding the AMA setup.sh. The Loop is running with AMA now, but still a little rough and I got a couple of questions. I built this rig (withour using this one) by just doing the normal steps, thereafter integrating auto-sens, MA and then changing to AMA. Then I started testing and running it with my son to see wether it works. When I invoke meal.json with this setup it seems to run twice 3 series of auto-sens. Thereafter it is reading the carbs from NS and is creating another meal.json. And additionally it is invoking auto-sens.json. From my point of view it seems that auto-sens is allready integrated when creating the meal.json. So running the auto-sens.json additionally is no longer necessary, right? Invoking the meal.json in my monitor-pump alias is also not necessary, if I run it after getting the carbhistory.json from NS, correct? So the only time to invoke meal.json is after getting the carbhistory.json? Thanks for clarification....
Paul Martin
@Spazholio
May 15 2016 18:18
Question about cron - does */5 mean “every 5 minutes” or “every 5th minute”? I’m noticing cron kick off at weird times, but expected it to only kick off at :05, :10, :15, :20, etc.
ddaniels1
@ddaniels1
May 15 2016 18:23
@scottleibrand regarding your post about making OpenAPS more accessible but worrying about the risks. Maybe there is a way to start a non profit to support the efforts and time of a small group of developers in order to take what is going on here to the level it could be accessible by everyone. The line has to be clear that people need to put the effort in and "do it themselves", but building out the OpenAPS toolkit to make it truly understandable and accessible to the average non programmer would be revoluntionary and is clearly not currently there. Sure everyone can keep revising the docs, but at some point time has to be reimbursed and it should be. No one would be looking to get rich, just make it possible. I would be interested in this in a serious way as I have had so many people already ask me for help when I barely have a hold on it at this point. As a physician and "non programmer" I may be able to provide important perspective. If you all are interested or have similar ideas we should talk.
Dana Lewis
@danamlewis
May 15 2016 18:26
Thx @ddaniels1. We are exploring a couple ideas with regard to scale.
Shane Mitchell
@Stavrosnco
May 15 2016 18:34
Hello everyone, I'm a newcomer working through documentation for the first time and just had a quick question. I have a Medtronic 551 and a 522. I see that the 522 is listed as compatible with openaps and the 551 is not. I just got to the point of using openaps to communicate with my pumps and I am able to query my 551 for it's settings, reservoir amount, model, etc. But am I accurate in understanding that the 551 is NOT supported because not all of the available commands work correctly?
ddaniels1
@ddaniels1
May 15 2016 18:43
For AutoSens Though the mean deviation output is changing the "new ISF" always stays the same as I run this throughout the day. IS that right? something configured wrong maybe?
.............................................................................................................................
p=0.60: 0.67, -1.02, 5.00
p=0.58: 0.43, -1.32, 4.50
p=0.56: 0.00, -1.67, 4.12
p=0.54: 0.00, -2.18, 3.76
p=0.52: -0.33, -2.61, 3.32
p=0.50: -0.67, -3.03, 2.76
p=0.48: -0.74, -3.56, 1.91
p=0.46: -1.00, -3.65, 1.62
p=0.44: -1.33, -3.98, 1.19
p=0.42: -1.33, -4.31, 0.92
p=0.40: -1.67, -4.72, 0.71
p=0.38: -1.67, -5.17, 0.30
p=0.36: -2.00, -5.54, -0.01
p=0.34: -2.33, -5.97, -0.33
p=0.32: -2.67, -6.30, -0.49
p=0.30: -3.17, -6.72, -0.68
p=0.28: -3.33, -7.08, -0.92
p=0.26: -3.67, -7.43, -1.22
Mean deviation: 3.93
Sensitivity within normal ranges
Basal adjustment 0.00U/hr
Ratio: 100%: new ISF: 89.0mg/dL/U
Dana Lewis
@danamlewis
May 15 2016 18:45
@Stavrosnco you can read data from most pumps, but it's a matter of whether you can issue temp basal commands. Easiest way to tell is to check and see if "pc connect" is present in the utilities menu. If it is, you won't be able to issue remote temps to the pump
Shane Mitchell
@Stavrosnco
May 15 2016 18:46
Thanks for the response, I missed that bit in the documentation!
Dana Lewis
@danamlewis
May 15 2016 18:47
:+1:
ddaniels1
@ddaniels1
May 15 2016 18:51
Assuming that 25-50% of the deviations should be negative and 50-75% should be positive, based on the above she is a little on the sensitive side? or is it because of fasting?
18 rows here not sure if that is hourly?
Hristina Dimova
@xpucuto
May 15 2016 18:54
@scottleibrand i will try to say with other words. :) When i suspended the pump on NS it appeard as 0U temp basal untill i resumed it. So setting a temp basal, before suspend is not necessary. If i am not mistaking this was the discussion about .
Dana Lewis
@danamlewis
May 15 2016 18:55
@xpucuto it is probably that the loop set the zero temp as you went to suspend
It is not programmed to do so
Unless you are already using mmhistorytools?
Hristina Dimova
@xpucuto
May 15 2016 18:56
this is a tricky question which answer i am not sure about
Dana Lewis
@danamlewis
May 15 2016 18:57
Which sensor are you using, Dexcom or Medtronic?
Scott Leibrand
@scottleibrand
May 15 2016 19:01
@ddaniels1 ignore column 2 and focus on column 4.
That shows that 36% are negative, which is within the normal range of 30-50%, so no adjustment to sensitivity is needed.
If you want ISF to be adjusted more aggressively, you'll want to upgrade to AMA.
ddaniels1
@ddaniels1
May 15 2016 19:28
@scottleibrand thanks!
Ben West
@bewest
May 15 2016 19:54
@ddaniels1 yes, I'm very concerned about ways to sustain the project
I've been doing this pro-bono for two full years now and cannot continue like this indefinitely
probably, not even the remainder of the year, we'll see
in terms of making it more accessible, there are several leads on that, I've got several GUIs that wrap everything up into a nice click-to-get-going-Apple/Facebook style experience
I hope in the past year, hopefully I've demonstrated many of the open source practices that allow community to grow and innovate safely together, make it modular, focus repos on small problems, make them work together, etc
Tia Ghose
@tiaghose_twitter
May 15 2016 23:50
Hi there, just wanted to pop in and say hello
My son was recently diagnosed with T1D and we're thinking of implementing an openAPS but not sure whether to do the Animas Ping (the pump we have) or buy an older minimed to implement the more documented version.
I had a question on whether someone is collecting glucose data from all these openAPS folks to see how it compares to their data prior to it? or if people are interested in that even?