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

21st
Jun 2016
Scott Leibrand
@scottleibrand
Jun 21 2016 00:05
I'd love to look at the details of that. feel free to PM me the screenshot and/or NS link
ryanchen26
@ryanchen26
Jun 21 2016 00:09
Will do @scottleibrand
eyim
@eyim
Jun 21 2016 00:10
@ryanchen26 Was there negative insulin on board at that point from low temping to recover from lows? That is when I see that behavior as well
ryanchen26
@ryanchen26
Jun 21 2016 00:14
@eyim Yes, the basal was zero (negative insulin) for awhile just before.
eyim
@eyim
Jun 21 2016 00:17
You might want to check what the net iob was. In our situation, his basal was set 0 for a couple of hours before he recovers so there is often -1 or -2 units iob. So that when he starts to recover, the algo thinks it needs to make that up so that he doesn't spike and will high temp him which will then push him low. In those times, I will turn off openaps.
Scott Leibrand
@scottleibrand
Jun 21 2016 00:18
looks like it basically high-temped just enough to get your IOB back up to zero
as far as it could tell, your BG was rising (so it expected that to continue), and net IOB was slightly negative, so it looked like a slight high temp was needed to counteract the rebound
what do you think caused the BG to drop again?
eyim
@eyim
Jun 21 2016 00:20
I think that is the 3rd or 4th person mentioning this scenario
Scott Leibrand
@scottleibrand
Jun 21 2016 00:20
the high temp itself couldn't have caused him to go low: that happened too fast
eyim
@eyim
Jun 21 2016 00:21
Ya probably not all due to the high temp - likely the body is in one of those states where it just keeps falling and needs more sugar to stay stable. The high temp just adds to the problem
Scott Leibrand
@scottleibrand
Jun 21 2016 00:22
and "low" in this case was only 85 :-)
eyim
@eyim
Jun 21 2016 00:22
Lol - I don't consider that a low :)
For us, it is generally exercise or heavy brain usage times (during finals) that causes this
Scott Leibrand
@scottleibrand
Jun 21 2016 00:23
I think this is just one of those cases where an insulin-only system simply can't do enough to prevent BG from dropping, so as long as you don't have glucagon as an option, the best an AP can do is bring BG back up over the course of a couple hours
Sarah Hodson Grady
@shgrady
Jun 21 2016 00:24
If i can iter_glucose, why can I not iter_glucose_hours....trying to add my second report
Scott Leibrand
@scottleibrand
Jun 21 2016 00:24
lots of people have seen issues with iter_glucose_hours lately. would be better to switch the docs to use oref0_glucose
eyim
@eyim
Jun 21 2016 00:26
Yep - I agree. No way for a insulin only system to stop this. I guess the question for me is whether to build into the algo something that accounts for this so that it won't high temp the person. Of course the trade off is that for the other 95% of the time and people, the high temp is appropriate. I was going to use temp targets and see if that helps us
Scott Leibrand
@scottleibrand
Jun 21 2016 00:26
@ryanchen26 you might also be interested to try autosens
yeah, setting a high temp target would be good when you suspect something like this is going on
@ryanchen26 looks like you've also had lots of periods of lengthy zero-temping, which might point to basals being too high during those periods. if you see that regularly, may want to adjust them downward: the loop will still be able to high-temp as needed, but will start from the lower base
or, if it's a temporary thing, autosens will help by adjusting basals downward and ISF upward across the board
ryanchen26
@ryanchen26
Jun 21 2016 00:29
are there docs on autosens yet?
Scott Leibrand
@scottleibrand
Jun 21 2016 00:29
I think that's in the ama-docs branch still
there may be something in phase 6 too, can't remember
ryanchen26
@ryanchen26
Jun 21 2016 00:30
Yeah we've had lots of zero basal...tuning has been a challenge.
Scott Leibrand
@scottleibrand
Jun 21 2016 00:31
still honeymooning?
ryanchen26
@ryanchen26
Jun 21 2016 00:32
Honeymoon is long over...we're just over a year in from the initial diagnosis.
Thanks for the notes, we'll adjust basal, and look into autosens.
Chris Oattes
@cjo20
Jun 21 2016 00:35
test the basal first :)
ryanchen26
@ryanchen26
Jun 21 2016 00:36
Done and done.
Sarah Hodson Grady
@shgrady
Jun 21 2016 00:54
I get the same no response.... with openaps use g4nonshare oref0_glucose
as I do with iter_glucose_hours 4
yet I can still iter_glucose 10
Sarah Hodson Grady
@shgrady
Jun 21 2016 01:27
it just puts brackets in front of command line
Paul Martin
@Spazholio
Jun 21 2016 01:34
That means it returned an empty array. So it ran, but just pulled no data.
What about if you specify —hours 2 after that command?
Andrew Dodd
@Entropy512
Jun 21 2016 01:42
@ryanchen26 By any chance, was this in the range of 3-5 hours after a meal bolus?
I think the Medtronic IOB estimation curve could be problematic with boluses
If you do a bit of crunching of the math with http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2769764/ (I should post one of my spreadsheets I've been playing with)
for a meal bolus that's 50% of total daily basal
Paul Martin
@Spazholio
Jun 21 2016 01:44
@scottleibrand @bewest You know how maxBasal is hardcoded to 3 x basal? Where is that set? Do you know offhand?
Andrew Dodd
@Entropy512
Jun 21 2016 01:44
at 3 hours you actually have a plasma insulin contribution from the bolus that is equal to the basal insulin concentration
at 4 hours it's around 30-50%
if your meal bolus was big (meal bolus total equal to daily basal) - you'll still have bolus contribution equal to your basal levels at 4-4.5 hours
Scott Leibrand
@scottleibrand
Jun 21 2016 02:35
@Spazholio search for maxSafeBasal.
@Entropy512 yeah, I was thinking @ryanchen26 might want to lengthen his DIA by an hour and see if that helps.
Andrew Dodd
@Entropy512
Jun 21 2016 02:39
Might help short-term, but that could have the side effect of mispredicting the peak and the magnitude of the peak. In the next week or two I'm going to de-cobweb my diffEq skills and grind out the math to get a more Nightscout-friendly implementation of the model I linked above.
What's tough is that predicts plasma insulin concentration - actual insulin activity is delayed compared to that, and there are multiple different models existing for that
but the common thing is an exponential decay
seems like a half-life of roughly an hour
Scott Leibrand
@scottleibrand
Jun 21 2016 02:40
I wonder if we could use CGM data to solve for the insulin activity curve directly.
(And pump data of course)
Jon Loeb
@biomuse
Jun 21 2016 02:40
That's using your noggin.
I love that more than i can say.
I have a feeling the watson approach to glucose prediction is going to be about something very much like this.
Andrew Dodd
@Entropy512
Jun 21 2016 02:41
It would be interesting, but the body sadly has enough variability that can be difficult.
That's why there seem to be a few different models for insulin activity
although the most widely-accepted one at this point seems to be the one used by the UVA/Padova simulator
Scott Leibrand
@scottleibrand
Jun 21 2016 02:42
Have you seen openaps/oref0#99 ?
Would love to get some people who actually understand the academic models working on helping make practical tools for using them to help people optimize their therapy.
Andrew Dodd
@Entropy512
Jun 21 2016 02:44
Yeah. I think one thing that will probably come out of it is a need to move away from Medtronic's very simple linear rise/linear fall activity model
Jon Loeb
@biomuse
Jun 21 2016 02:44
@Entropy512 dunno. once healthkit is in wide use weight, activity will be associated with data, no? Intraindividual noise due to site variation will be there of course but that can be cleaned.
Andrew Dodd
@Entropy512
Jun 21 2016 02:45
@biomuse I'm talking less about site variation than time variation of metabolism
I've seen my ISF vary daily by at least a 2:1 ratio
and I've had days/nights (like last night) where I seem to have a quarter the ISF of normal
Jon Loeb
@biomuse
Jun 21 2016 02:45
ditto. but that must be correlated with something.
Scott Leibrand
@scottleibrand
Jun 21 2016 02:45
Even if not, autosens can detect it.
With openaps/oref0#98 we can probably get detection times down to a few hours.
Andrew Dodd
@Entropy512
Jun 21 2016 02:46
I think autosens can detect the anomalies, but automatically detecting some of the model parameters might be challenging
Scott Leibrand
@scottleibrand
Jun 21 2016 02:46
@Entropy512 would love your comments on 99
Andrew Dodd
@Entropy512
Jun 21 2016 02:46
The problem is there seem to be very few models/information on modeling fat
I think the first thing on 99 is to consider a model for insulin action that doesn't have Medtronic's linear rise/fall of activity - It's probably OK for basal adjustments since any temps will decay to be negligible within the DIA period
but I think where things can get complicated is with boluses
Jon Loeb
@biomuse
Jun 21 2016 02:48
well translipid diffusion of subcutaneously delivered time-critical hormones is pretty much an issue limited to... us ;(
Andrew Dodd
@Entropy512
Jun 21 2016 02:48
in those cases, being 1-2 more half-lives longer could be significant
I'll try to clean up one of my spreadsheets/implementations of a subset of the model tomorrow
Matt Pressnall
@logichammer
Jun 21 2016 02:50
@sulkaharo , @scottleibrand , @jasoncalabrese, updated AMA and now no more dupes / over-estimation of carbs! I have seen the "Removing duplicate" text that I saw in Jason's github note but I can definitely say we aren't seeing the doubling of carbs that we were seeing before so thanks!!!
Meant to say "I haven't seen the ..."
Andrew Dodd
@Entropy512
Jun 21 2016 02:51
Most notably, it introduces some nonlinear differential equations into the model
which means you can't solve it without doing a numerical integration
e.g. stepping the model by a minute at a time for example
and that could get CPU-consuming for something like Nightscout on Azure
Jon Loeb
@biomuse
Jun 21 2016 02:55
Though maybe not prohibitive for the iPhone.
Andrew Dodd
@Entropy512
Jun 21 2016 03:10
Probably not an issue for a standalone system like anything running openaps
OK, what's everyone's preferred method for putting up an ODS spreadsheet here? :)
hmm lemme see if Google Docs handles the chart portion properly
Andrew Dodd
@Entropy512
Jun 21 2016 03:21
OK, try playing with this - you need LibreOffice or OpenOffice
you can use Google Sheets to view it but there are no horizontal axis gridlines
Right now I've mangled it to eliminate the Kd factor which is dose-dependent
I have not figured out how to model that for continuous infusion
Compares the plasma insulin concentrations of a 10U bolus vs. steady-state 20U/day basal
note that in reality, that bolus insulin concentration would be on top of the basal
you can see that at 3.5h for "typical" model values, the plasma concentration contribution from the bolus is equal to the basal contribution
e.g. you've got double your basal insulin concentration there
at 4.5 hours it's roughly 50% of basal
or if it were a 20U bolus
it would equal basal at 4.5h
actual activity of the insulin is delayed compared to plasma concentration
Andrew Dodd
@Entropy512
Jun 21 2016 03:26
there are at least 3 different models I've seen for that...
Paul Martin
@Spazholio
Jun 21 2016 03:36
I just noticed something interesting - except for the order of a couple items, my pumphistory.json and my pumphistory-zoned.json are identical.
Well, at least on my Pi3 rig. My Edison seems to be doing stuff right. That’s weird.
Andrew Dodd
@Entropy512
Jun 21 2016 03:40
man I wish the t:slim had some provision for remote control - that looks like a really nice pump
Also, based on what I'm reading in the openaps docs - the Animas Ping does not have a provision for remotely setting a temp basal?
only remote bolus?
what the hell were they thinking?
Scott Leibrand
@scottleibrand
Jun 21 2016 03:45
It may exist, but it's not used by anything and we don't have any tools for discovering it.
Andrew Dodd
@Entropy512
Jun 21 2016 03:46
Seems like a big oversight to not have basal settings in the remote
Scott Leibrand
@scottleibrand
Jun 21 2016 04:03
There isn't a single pump maker who does afaik.
Andrew Dodd
@Entropy512
Jun 21 2016 04:19
Other than OmniPod since that's 100% remotely controlled
time to sleep
Scott Leibrand
@scottleibrand
Jun 21 2016 04:22
:+1:
Sulka Haro
@sulkaharo
Jun 21 2016 04:48
@logichammer great
C-Ville
@C-Ville
Jun 21 2016 04:58
The AC Combo does allow to access all settings (including temp basal) with its remote, but the pump is not so wide spread in the US and the communication seems to be highly encrypted
ryanchen26
@ryanchen26
Jun 21 2016 05:02
@Entropy512 yes, 4.5 hours after lunch. He went high to 280, was high temping and not budging, given a 0.2u correction bolus, dropped fast, caught it around 80, 4g glucose, rise to 128, high temp kicks in, drops again, 2g glucose slows to 90, climbs a bit, we blood check before dinner, dexcom says 120, bg check is actually 63. So many variables...plus we changed the dexcom sensor this morning... for the last 3hrs the bg has been rising steadily despite the high temp kicking in and a conservative bolus...we're working on it.
ryanchen26
@ryanchen26
Jun 21 2016 05:12
@scottleibrand changed active insulin time from3 to 4 hrs.
Andrew Dodd
@Entropy512
Jun 21 2016 12:21
@C-Ville Yeah I've chatted with timomer about the Accu-Chek - looks interesting but as you say rare in the US
and looking at the manual - maybe I'm misreading it but it only allows 0.1u/hr basal increments???
seems like insufficient resolution for pretty much all diabetics
Andrew Dodd
@Entropy512
Jun 21 2016 12:49
OK their manual is deceptive
.01u/h up to 1 u/h
Sarah Hodson Grady
@shgrady
Jun 21 2016 13:25
@Spazholio I tried -hours 1 last night, still nothing, but thanks for the info that it was indicating an empty array.
Andy Probolus
@AndyProbolus_twitter
Jun 21 2016 15:50
hello, I'm trying to make documentation changes to submit to you in the future, but I'm having trouble with the line endings showing up as changes. Don't want to submit them all to you, what line ending is the project using? I'll try to match
Rim Dargis
@rjd107
Jun 21 2016 15:54
Has anyone had any luck finding a replacement WirelessThings ERF module since the link is broken and i dont see "ERF" anywhere on the page? I'm looking to build the Intel Edison with ERF stick in the US..
GabeA23
@GabeA23
Jun 21 2016 16:24
This message was deleted
Dana Lewis
@danamlewis
Jun 21 2016 16:45
@AndyProbolus_twitter what do you mind ,line endings? At this point, I’d suggest just submitting a PR and I can take a look - very easy to then edit the PR if there’s anything wonky, but chances are it’ll be fine!
@rjd107 if I remember correctly they’ve stopped selling those? have you considered a TI? That’s a good alternative to carelink with a much better range
Neal
@tnharvey
Jun 21 2016 16:52
Hi, I'm currently using an Animas Vibe (I know, not currently openAPS, not even in dev since no commands have been captured). I've got a 722 that I want to start openAPS on in a month or two here, but my G4 sensor just died, and I now need to decide between replacing with a G4 or upgrading to the G5. I'm on Android and not wanting to switch to iOS, but I need something I can make work until I can dedicate myself to getting openAPS going. If there's a working solution to connecting the G5 to my android (via openAPS maybe?), I can do that too. Or if anyone knows of a Dexcom Beta I can work with in the meantime. SO! The question is, knowing I want to do openAPS with the 722 in a couple months on Android, but need access to my data now, G4 or G5?
Dana Lewis
@danamlewis
Jun 21 2016 16:55
@audiefile ^ xdrip best option? or other ideas?
Neal
@tnharvey
Jun 21 2016 17:34
This message was deleted
@tnharvey sorry, G4 *transmitter (not sensor)
Ben West
@bewest
Jun 21 2016 17:40
I favor G4, I like have a receiver that can talk wireless
GabeA23
@GabeA23
Jun 21 2016 17:48
Hey all. I'm in Phase 1 of my openaps setup and I can't seem to get the openaps plugin enabled on my Azure NS website. I added the word "openaps" to my list of enabled plugins on Azure, but when I go to my NS url and check under settings, "openAPS" does not show up in the list of available plugins.
Paul Martin
@Spazholio
Jun 21 2016 17:53
@bewest In pumphistory.json - what format is the _date field in? It doesn't appear to be UNIX time...
GabeA23
@GabeA23
Jun 21 2016 17:53
openaps error.png
Paul Martin
@Spazholio
Jun 21 2016 17:56
I'm toying with the notion of taking the pumphistory.json file that I get every 5 mins, and pushing the new data into a sqlite DB so that hopefully I can stop having to pull a pull pumphistory-24.json. It's basically just a thought experiment at this point, but I think it could yield some interesting speed/efficiency increases.
I'm trying to find a field in each JSON entry in pumphistory.json that can use as a unique field - one that's present in each entry, but never duplicated. So far, I don't think it's possible without pulling in another field to check against.
Ben West
@bewest
Jun 21 2016 18:12
dateString or timestamp is what you should use
the _date is the raw binary for the date segment as it appeared in raw pump memory
I would be surprised if introducing (another) db made things faster
Paul Martin
@Spazholio
Jun 21 2016 18:13
pumphistory.json doesn't have dateString. And timestamp is duplicated across two entries.
I can dump everything into a sqlite DB in less than 2s. Meaning, I can pull out 24h worth of stuff to a JSON file VERY quickly, rather than waiting for it to get pulled from the pump.
Ostensibly, we wouldn't need to pull 24h of records from the pump anymore. For me, that's EASILY the longest pull I have to do. And the one most prone to dying.
Ben West
@bewest
Jun 21 2016 18:15
right, so we've talked about doing --append mode for report
Paul Martin
@Spazholio
Jun 21 2016 18:16
That could be workable too. I can see some potential issues with that as well, but those are fairly easily mitigated, I believe.
Ben West
@bewest
Jun 21 2016 18:17
you still have to talk to the device at some point to get new records... these devices all have a minimum "data-transfer" size, in which the least amount of data you ever get is a full page of data
Paul Martin
@Spazholio
Jun 21 2016 18:17
pumphistory.json goes pretty quickly. pumphistory-24h.json however, seems to take quite a bit longer.
I'm running this past you since I consider you to be the SME here. =)
Ben West
@bewest
Jun 21 2016 18:18
my advice is to do something incredibly simple and easy to debug and not worry about prematurely optimizing things
what you want has nothing to do with needing a db
you want to append just new stuff to a full cached version of prior stuff
Paul Martin
@Spazholio
Jun 21 2016 18:19
My reasoning behind using a DB is that you can enforce uniqueness without iterating over an array, whereas you can't really with a flat file. But I absolutely see what you mean.
Ben West
@bewest
Jun 21 2016 18:19
lsgaps can examine your cache, assuming one exists
what's missing is the append part and the ability to use a gap spec as the definition for "how much data to grab"
with those two things you could do report add updated.json --append JSON pump iter_pump --gap recent-missing.json
two features rather than introducing db
Paul Martin
@Spazholio
Jun 21 2016 18:22
And currently, the --append function is theoretical, right? It's not present in the codebase yet?
Ben West
@bewest
Jun 21 2016 18:22
is it so hard to imagine?
it's a relatively easy feature to add, simpler than adding a db, for example ;-)
Paul Martin
@Spazholio
Jun 21 2016 18:22
How do you mean?
I mean, I can't USE this today, right? =)
Ben West
@bewest
Jun 21 2016 18:23
it's as much a theory as the db, sure
Paul Martin
@Spazholio
Jun 21 2016 18:23
Point taken. =)
Ben West
@bewest
Jun 21 2016 18:23
it has my support and would work for things other than pump history
it would improve all the other features as well, for example
so it'd work for both entries and treatments
not super hard to think about implementing an append tool quickly using cat * | json -g in the meantime
the --gap feature has been spread to several dexcom uses, if someone wants to help port that logic over to the medtronic uses, we'll have half of it done
part of it, is you exhibited interest in entries first, so I've been working on that
there are obviously lots of features that would be nice to add, it's easier to ask for more faster than I can write them
as a result dexcom uses now support --gaps spec
Paul Martin
@Spazholio
Jun 21 2016 18:46
I get that, which is why I was going to try taking a stab at it before asking you to do it for me. I'm still getting my head around the code flow of all the Python and JS that this stuff is written in. Following what a bit of code is doing is simple. Finding what fed into and what that code feeds into is more difficult.
Jarred Yaw
@jyaw
Jun 21 2016 18:54
@scottleibrand is there a grand master list of models (for use in the discussion of #99) being used in the area of closed loop research... maybe one you guys are maintaining? or is google my best option?
Dana Lewis
@danamlewis
Jun 21 2016 18:57
You mean all models including academic? Yes, use google @jyaw I don't know anyone who's tracking that open source
Also keep in mind the best model is one we can get done :) without taking a year to do it
@GabeA23 which version of NS are you on?
GabeA23
@GabeA23
Jun 21 2016 19:01
@danamlewis 0.8.4
that should be the most up-to-date version, no?
Dana Lewis
@danamlewis
Jun 21 2016 19:04
No
Jarred Yaw
@jyaw
Jun 21 2016 19:04
@danamlewis ya, academic. Totally agree on the "we can get done" part. Was wanting to understand what the gap was between the linear approach we use in OpenAPS and what has been looked at in academia.
Dana Lewis
@danamlewis
Jun 21 2016 19:05
@jyaw :) would be good if once you do some sleuthing to share that in 99 for some context. Also btw check your Twitter DMs in case you haven't already
Jarred Yaw
@jyaw
Jun 21 2016 19:05
will do
*under NS setup heading
GabeA23
@GabeA23
Jun 21 2016 19:07
Thanks @danamlewis
So should I be using the docs at Github or http://openaps.readthedocs.io/ ?
Also, my repo is up to date according to the "update my fork" tool: http://nightscout.github.io/pages/update-fork/
Dana Lewis
@danamlewis
Jun 21 2016 19:16
You can look either place for docs. RTD is slightly easier to nav for many but it's personal preference . But content is same
Need dev
Rachel Sandlain
@audiefile
Jun 21 2016 19:21
@tnharvey there isn't yet an Android solution for g5. There is a beta being tested. I'm not sure how it's going/how close to release/how open it is. I know they were initially keeping it small because of hardware differences. I'd search/ask in the cgm in the cloud group for an update. Personally, I'm a fan of the g4
@GabeA23 after you update to dev, I'd also add careportal to your enable. You're going to need it enabled to see the basal display
GabeA23
@GabeA23
Jun 21 2016 19:25
@danamlewis It works. You are a wizard. Thanks!!!!!!
@audiefile Thanks! will do.
Dana Lewis
@danamlewis
Jun 21 2016 19:26
@GabeA23 woohoo! glad it resolved. and thanks @audiefile. may also want to check out that example in the docs I linked to for other things listed in addition to careportal you may want to ultimately add
ryanchen26
@ryanchen26
Jun 21 2016 19:36
Autosens reads like it throttles the ISF number based on various inputs. Is there a range limit ( how high can it go?) for this feature of dynamically adjusting the ISF, or am I not understanding it correctly?
Andy Probolus
@AndyProbolus_twitter
Jun 21 2016 19:52
@danamlewis check out various updates to documentation based on our 'newbie' experiences #223, just committed a few small PRs with a ton of changes that are really whitespace due to the line ending issue we are having
Dana Lewis
@danamlewis
Jun 21 2016 19:54
@ryanchen26 we’ve been discussing a range limit but don’t think we’ve enabled one yet. but sounds like you’d +1 that addition? Do you seem extremes on both sides?
@AndyProbolus_twitter excellent, thanks for PRing those!
Andy Probolus
@AndyProbolus_twitter
Jun 21 2016 19:55
@danamlewis feedback is appreciated if there are other ways you want stuff submitted, want to be helpful! thanks for all you do.
Dana Lewis
@danamlewis
Jun 21 2016 20:02
:+1: awesome @AndyProbolus_twitter just made some comments about a couple of different paths forward depending on time/interest!
Garrett Webb
@garetis
Jun 21 2016 20:03
When I get this report, what is being compared?
pi@raspberrypi:~/openaps2 $ openaps report invoke enact/suggested.json
determine-basal://text/shell/enact/suggested.json
BG data is too old, or clock set incorrectly Tue Jun 21 2016 14:49:49 GMT-0500 (CDT) vs Mon Jun 20 2016 03:11:51 GMT-0500 (CDT)
reporting enact/suggested.json
Paul Martin
@Spazholio
Jun 21 2016 20:04
Your CGM/NS time versus your system time.
Are you running an older version of openaps?
Garrett Webb
@garetis
Jun 21 2016 20:05
0.1.5
Paul Martin
@Spazholio
Jun 21 2016 20:06
Hmmm, ok.
Garrett Webb
@garetis
Jun 21 2016 20:07
pi@raspberrypi:~/openaps2 $ openaps status-upload
ns-status://JSON/shell/monitor/upload-status.json
Could not require: /home/pi/openaps2/enact/suggested.json [SyntaxError: /home/pi/openaps2/enact/suggested.json: Unexpected end of input]
Could not require: /home/pi/openaps2/enact/enacted.json { [Error: Cannot find module '/home/pi/openaps2/enact/enacted.json'] code: 'MODULE_NOT_FOUND' }
Could not require: /home/pi/openaps2/monitor/status.json { [Error: Cannot find module '/home/pi/openaps2/monitor/status.json'] code: 'MODULE_NOT_FOUND' }
reporting monitor/upload-status.json
^CTraceback (most recent call last):
  File "/usr/local/bin/openaps", line 4, in <module>
    __import__('pkg_resources').run_script('openaps==0.1.5', 'openaps')
  File "build/bdist.linux-armv7l/egg/pkg_resources/__init__.py", line 719, in run_script
  File "build/bdist.linux-armv7l/egg/pkg_resources/__init__.py", line 1504, in run_script
  File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/EGG-INFO/scripts/openaps", line 169, in <module>
    app( )
  File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/cli/__init__.py", line 51, in __call__
    self.run(self.args)
  File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/EGG-INFO/scripts/openaps", line 154, in run
    builtins.dispatch(args, self)
  File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/builtins.py", line 48, in dispatch
    get_alias(args.command, app)(args)
  File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/builtins.py", line 34, in __call__
    exit(call(cmd + args.args))
  File "/usr/lib/python2.7/subprocess.py", line 522, in call
    return Popen(*popenargs, **kwargs).wait()
  File "/usr/lib/python2.7/subprocess.py", line 1384, in wait
    pid, sts = _eintr_retry_call(os.waitpid, self.pid, 0)
  File "/usr/lib/python2.7/subprocess.py", line 476, in _eintr_retry_call
    return func(*args)
KeyboardInterrupt
I don't know if that's relevant of not
Also, I get "no recent treatments to upload" when I run openaps upload-recent-treatments.
Does it need a certain amount of information before it will upload (I just got my dexcom today, and it's started , but only for less than an hour of bg data)
Gera Yeremin
@gera3d
Jun 21 2016 20:13
Hi everyone so I got all my parts and my pump is ready to go.
I wanted to see what you guys think of this idea.
David Cintron
@loudestnoise
Jun 21 2016 20:13
:clap:
Gera Yeremin
@gera3d
Jun 21 2016 20:14
What if I twich stream myself putting everything together
twitch
Just unboxing everything and getting it ready for the software
Sarah Hodson Grady
@shgrady
Jun 21 2016 20:15
ha ha, I unboxed everything as it came...what patience you have
Gera Yeremin
@gera3d
Jun 21 2016 20:16
Thanks @Spazholio I am just seeing this now. That is very generous of you. I already secured one though.
Paul Martin
@Spazholio
Jun 21 2016 20:17
@gera3d No worries - glad it all worked out. =)
Gera Yeremin
@gera3d
Jun 21 2016 20:17
another quick question for the group between the 515 and the 522 what pump should I use for my 5 year old?
Currently we are using the newer Animus Ping with the remote.
Sarah Hodson Grady
@shgrady
Jun 21 2016 20:18
@gera3d @Spazholio wondering if this is about the CareStick? There was somebody on here who had a pump they were not sure if was compatible and was hesitant to buy a CareStick before using it to find out. I will look for them if thats what that was about?
Gera Yeremin
@gera3d
Jun 21 2016 20:20
@shgrady Yes it was about the CareStick. I actually found someone on craigslist that just sent me one. Generous community.
Sarah Hodson Grady
@shgrady
Jun 21 2016 20:21
Ok, somebody on here did not have the $60 bucks to spend on a CareLink in order to check out if the pump they have was compatible, ugh hard to search on here, I remember Ben West responding to it.
Neal
@tnharvey
Jun 21 2016 20:27
@audiefile I'll go with G4 then and start fiddling in a couple months, thanks for your input on this! and thanks @danamlewis for getting me connected on here!
ryanchen26
@ryanchen26
Jun 21 2016 20:27
@danamlewis Well, no range limit could be a good thing. When we give BG correction boluses we're usual calculating it with an ISF of 300. I'm more just wondering what the system can do at this point. Thanks.
Dana Lewis
@danamlewis
Jun 21 2016 20:28
@tnharvey welcome!
@gera3d you’ll find the 522 a smidge easier, but they’ll both work
the form factor is the same, so the only difference is a slightly easier setup process for you
Gera Yeremin
@gera3d
Jun 21 2016 20:29
@danamlewis What about features?
522 got it.
Dana Lewis
@danamlewis
Jun 21 2016 20:29
look on the hardware page, it says which of the models (it’s the x12 or the x15) that is slightly limited. but yea, x22 is best on all fronts
Gera Yeremin
@gera3d
Jun 21 2016 20:30
ok I will check that out tonight
Garrett Webb
@garetis
Jun 21 2016 20:32
So openaps upload-recent-treatments is the way to upload data from the rig to NS?
Garrett Webb
@garetis
Jun 21 2016 20:39
Also, does openaps gather redundantly invoke monitor/iob.json twice?
Scott Leibrand
@scottleibrand
Jun 21 2016 20:53
@ryanchen26 @danamlewis currently we have a +/- 2x limit on autosens ratio. I've been thinking about reducing it to +/- 1.5x
I think the 515 is almost identical to the 522 for our purposes. it's the 512 that is missing a bunch of stuff. 522 is still slightly better though, all else being equal.
Rachel Sandlain
@audiefile
Jun 21 2016 21:10
I used a 712 for 10+ months. It will work but it's fiddly. You have to generate the settings files by hand
Paul Martin
@Spazholio
Jun 21 2016 21:34
Are there any docs for setting up a Pebble watch? I’ve got the iPhone app setup, but in my loop, it’s supposed to be doing SOMETHING with the Pebble, and I have no idea how to implement that. =)
Dana Lewis
@danamlewis
Jun 21 2016 21:37
Have you looked at urchin yet @Spazholio ?
It's pretty straight forward
Paul Martin
@Spazholio
Jun 21 2016 21:37
That’s exactly what I have.
It’s working pretty well. But…what’s my loop doing with regards to Pebble? I’ve never set it up, it was just kinda….there. =)
Andrew Dodd
@Entropy512
Jun 21 2016 21:43
@bewest @tnharvey Yeah, I tried the G5 with an iPod Touch, glad I didn't upgrade my receiver so I could go back to the G4. G4 transmitter range is vastly superior to the G5, plus its broadcast nature means there are a lot of aftermarket receiver solutions people have cooked up. (I only use G4+ Share so far, but it's nice knowing you have other options)
Neal
@tnharvey
Jun 21 2016 21:50
@Entropy512 Thanks, that's what I was looking to hear was whether the G5 was worth grabbing an old iPhone or not, with the hope of less hardware later on, but from you and Rachel said, there's a lot more support for the G4, and not enough development to merit the G5 headache. Maybe in a year =]
Andrew Dodd
@Entropy512
Jun 21 2016 21:52
My opinion would probably be different if it weren't for the fact that the "default" G5 setup has horrible range and no EGV backfilling
I believe someone found a backfill command somewhere
but I don't know if anyone ever got that to work, pretty blah that the default app doesn't support it
tons of missed readings
Neal
@tnharvey
Jun 21 2016 21:55
@Entropy512 No kidding, I've had periodic range issues with the G4, I can't imagine upgrading to have more range problems, no thanks. And I read about the EGV backfill issue, for all the foot-dragging the FDA has done with the Android version, you'd think they'd have sorted an issue like that out...
Andrew Dodd
@Entropy512
Jun 21 2016 21:56
Some of the issues COULD be just the iTouch sucking, but... Since it's either iTouch working well or irreversibly "upgrading" my receiver (which, as I understand it, removes Share...) - went back to G4
also G5 Tx are hardcoded to die a bit past 3 months no matter what the battery state
G4 Tx batteries seem to last WELL beyond their 6-month rating on average
Matt Walker
@walker0
Jun 21 2016 21:57
@Entropy512 The G5 receiver is also approximately the brightness of the sun
Andrew Dodd
@Entropy512
Jun 21 2016 21:57
I'm close to 12 months and still haven't gotten a lowbat alarm
@walker0 Hardware change, or does that apply also to G4 receivers with the firmware upgrade?
Neal
@tnharvey
Jun 21 2016 21:58
yeah, I'm at I think about 14 months now on my G4 and it finally died, I don't like the hardcoded lifespan thing that I'm starting to see with some devices now
Matt Walker
@walker0
Jun 21 2016 21:58
I believe both, they made the graph background white an don't have reasonable adjustable backlight settings so you end up blinding yourself at night
Garrett Webb
@garetis
Jun 21 2016 22:01
Does anyone know how to resolve syncing the CGM/NS time and System time, so as to not have time errors when running enact/suggested.json?
Andrew Dodd
@Entropy512
Jun 21 2016 22:07
@walker0 OK, so they made the Rx behave like the iOS client? vomit
Marianne Smith, PMP
@mariannesmith77_twitter
Jun 21 2016 22:08
hi everyone, working on reinstalling reports on a second sd card. Trying to add tz timezone
I got this sequence : pi@raspberrypi:~/myopenaps $ openaps device add tz timezones
Traceback (most recent call last):
File "/usr/local/bin/openaps-device", line 4, in <module>
import('pkg_resources').run_script('openaps==0.1.5', 'openaps-device')
File "build/bdist.linux-armv7l/egg/pkgresources/_init.py", line 719, in run_script
File "build/bdist.linux-armv7l/egg/pkgresources/_init.py", line 1504, in run_script
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/EGG-INFO/scripts/openaps-device", line 35, in <module>
app( )
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/cli/init.py", line 47, in call
self.configure_parser(self.parser)
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/EGG-INFO/scripts/openaps-device", line 21, in configure_parser
self.commands = devices.configure_commands(parser, parent=self)
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/devices/init.py", line 36, in configure_commands
parent.devices = get_device_map(parent.config)
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/devices/init.py", line 21, in get_device_map
for device in Device.FromConfig(vendors, conf):
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/devices/device.py", line 76, in FromConfig
name = json.loads(candidate.split(' ').pop( ))
File "/usr/lib/python2.7/json/init.py", line 338, in loads
return _default_decoder.decode(s)
File "/usr/lib/python2.7/json/decoder.py", line 366, in decode
obj, end = self.raw_decode(s, idx=_w(s, 0).end())
File "/usr/lib/python2.7/json/decoder.py", line 384, in raw_decode
raise ValueError("No JSON object could be decoded")
ValueError: No JSON object could be decoded
can you tell me what JSON object it is looking for?
Matt Walker
@walker0
Jun 21 2016 22:22
@Entropy512 yep, I only switched because the G5 is officially approved for dosing in Australia
Andrew Dodd
@Entropy512
Jun 21 2016 22:23
Interesting... To my knowledge, there is no real difference between G4 and G5 in terms of accuracy, etc. After all, they both use the same sensor
tbh that's a MAJOR flaw in Dexcom - I can understand not giving the raw data
but not making the "noise" determination readily available is a gigantic WTF moment
the "noise" determination is effectively a self-diagnostic
it tells you when something is wrong
I cannot imaging G5 being approved for dosing without notifying the user that the sensor switched from "Clean" to "Light" filtering
because I have multiple times seen my Dexcom switch to "light" after a carb correction
telling me that my bloodsugar had started dropping again because it was rising too fast!
Scott Leibrand
@scottleibrand
Jun 21 2016 22:32
I think they fixed that downtick on the switch to filtered when they went to the 505 firmware
Matt Walker
@walker0
Jun 21 2016 22:37
@Entropy512 agreed, although I find the G5 quite accurate most of time except on the first day when it constantly tries to tell me I'm low
Andrew Dodd
@Entropy512
Jun 21 2016 22:38
@scottleibrand Definitely not, as my G4 is well post-505 (started last July), and the G5 is clearly post-505
Paul Martin
@Spazholio
Jun 21 2016 22:38
There’s no reason the Pebble BLE and the Dexcom BLE should be causing any kind of interference with one another, is there?
Andrew Dodd
@Entropy512
Jun 21 2016 22:39
They shouldn't, there's enough channel hopping and timespacing that it shouldn't be a problem
that's another thing that was "wtf?" about the G5
a system with all of those retransmissions and timeouts should not have been that much of an unreliable nightmare
G5 BLE advertisements are far more numerous than the actual G4 broadcasts
yet the system is less reliable and has half the battery life despite larger battery
Paul Martin
@Spazholio
Jun 21 2016 22:41
Well, I just started using a Pebble, and I’m noticing that the Dexcom seems to lose the BT icon fairly regularly. It recovers, naturally, but it’s gone, and instead of NS showing a time of <5 mins for syncing with share, it’s generally showing every 10 mins. Just odd is all.
Andrew Dodd
@Entropy512
Jun 21 2016 22:44
hmm, interesting
I'm using a Moto 360 and rarely see a sync time of >7min
despite newer versions of Android (and xDrip's workarounds for them) being a headache
overall most of my "stops updating until phone reboot" problems disappeared with xDrip+
had one day of issues last week, but lately - no problems
Paul Martin
@Spazholio
Jun 21 2016 22:47

And I’m sure this is unrelated, but my loop just started failing with this:

pump://JSON/iter_pump_hours/monitor/pumphistory.json
Traceback (most recent call last):
  File "/usr/local/bin/openaps-report", line 82, in <module>
    app( )
  File "/usr/local/lib/python2.7/dist-packages/openaps/cli/__init__.py", line 51, in __call__
    self.run(self.args)
  File "/usr/local/bin/openaps-report", line 75, in run
    output = app(args, self)
  File "/usr/local/lib/python2.7/dist-packages/openaps/cli/subcommand.py", line 52, in __call__
    return self.method.main(args, app)
  File "/usr/local/lib/python2.7/dist-packages/openaps/reports/invoke.py", line 40, in main
    output = task.method(args, app)
  File "/usr/local/lib/python2.7/dist-packages/openaps/uses/use.py", line 45, in __call__
    output = self.main(args, app)
  File "/usr/local/lib/python2.7/dist-packages/openaps/vendors/medtronic.py", line 639, in main
    for rec in self.range( ):
  File "/usr/local/lib/python2.7/dist-packages/decocare/models/__init__.py", line 67, in __call__
    for record in page:
  File "/usr/local/lib/python2.7/dist-packages/decocare/models/__init__.py", line 46, in download_page
    for record in self.find_records(page):
  File "/usr/local/lib/python2.7/dist-packages/decocare/models/__init__.py", line 157, in find_records
    decoder = history.HistoryPage(response.data, self.inst)
  File "/usr/local/lib/python2.7/dist-packages/decocare/history.py", line 777, in __init__
    if lib.BangInt(crc) != computed:
  File "/usr/local/lib/python2.7/dist-packages/decocare/lib.py", line 235, in BangInt
    ( x, y ) = ints
ValueError: need more than 0 values to unpack

It’s the exact same error - the BangInt one - each time. It’s happened about 8-9 times. Anyone ever seen that or know what’s up?