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

16th
Sep 2016
Tim Mecklem
@tmecklem
Sep 16 2016 00:08
possible jackpot… http://blog.thiseldo.co.uk/?p=874
(the part about adding an antenna at least)
Tim Mecklem
@tmecklem
Sep 16 2016 00:33
So it looks like several sources are claiming an 82mm wire soldered to the antenna pin should do the trick.
Assuming (with very limited Computer Sciency hardware background and whatever I can remember from studying for my ham license), this is a 1/4 wave wire whip antenna for a frequency ~915MHz
Looks like me and my trusty calipers and soldering iron are going on an adventure together.
Eric
@ecc1
Sep 16 2016 00:35
One tip: cut the wire longer than needed, solder it, then measure and cut it to exact quarter-wavelength once it's attached.
Tim Mecklem
@tmecklem
Sep 16 2016 00:36
That’s great advice! Thank you @ecc1.
Any tips for wire gauge?
Eric
@ecc1
Sep 16 2016 00:40
I'm using 26 gauge so I can bend it easily inside a case, and it seems to work OK (this is on a RFM69 module, but same idea, just no need to disconnect an on-chip antenna because it doesn't have one).
garykidd
@garykidd
Sep 16 2016 00:41
@scottleibrand
Eric
@ecc1
Sep 16 2016 00:41
I'm assuming you're aiming for as small a rig as possible. If you want to optimize range, you can add a jack (u.FL or SMA) for an external antenna.
garykidd
@garykidd
Sep 16 2016 00:42
Wondering if it's easy enough to assign a different port to ti stick to test if it had anything to do with a specific port.
Tim Mecklem
@tmecklem
Sep 16 2016 00:45
I’m trying for a smaller rig for sure.
It’ll likely be within several feet of the pump the vast majority of the time.
Ryan Dalisky
@rdalisky
Sep 16 2016 00:49
Hi all. I am working on a new install. I have a 722 with enlite connected. I'm struggling with these instructions for correctly pulling BG data: https://openaps.readthedocs.io/en/dev/docs/walkthrough/phase-2/Using-oref0-tools.html#cleaning-cgm-data-from-minimed-cgm-systems
I can get my bgs and format it using mm-format-ns-glucose, but oref0 doesn't seem to recognize it
Tim Mecklem
@tmecklem
Sep 16 2016 00:50
@rdalisky I just did an enlite setup a few days ago
Ryan Dalisky
@rdalisky
Sep 16 2016 00:51
Sweet!
Any tips?
Tim Mecklem
@tmecklem
Sep 16 2016 00:52
I don’t think oref0 wants to see the mm-format-ns-glucose output. That’s specifically to be run against the raw glucose to report to nightscout
oref0 wants to see the glucosetools clean output
Ryan Dalisky
@rdalisky
Sep 16 2016 00:52
ahhh
Tim Mecklem
@tmecklem
Sep 16 2016 00:53
So as I understand it, you’ve got the iter_glucose report putting out a “raw” file.
You run the glucosetools clean against that to get input for the oref0 looping, and you mm-format-ns-glucose against the raw for a format that nightscout can ingest.
I think I may have also needed to run it through tz for timezone information to be populated before oref0. I can’t remember and it was for a friend so I don’t have the looping device here.
Ryan Dalisky
@rdalisky
Sep 16 2016 00:56
That helps a bunch, thank you
garykidd
@garykidd
Sep 16 2016 00:57
@scottleibrand @jasoncalabrese so i'm taking it that rf_enable field relates to meter option? and paradigm relates to remote option? within settings/settings.json......
pi@raspberrypi:~/APS301 $ cat settings/settings.json
{
  "low_reservoir_warn_point": 25, 
  "keypad_lock_status": 0, 
  "rf_enable": false, 
  "temp_basal": {
    "type": "Units/hour", 
    "percent": 100
  }, 
  "low_reservoir_warn_type": 0, 
  "insulinConcentration": 100, 
  "audio_bolus_enable": true, 
  "variable_bolus_enable": true, 
  "alarm": {
    "volume": 3, 
    "mode": 2
  }, 
  "maxBasal": 2.0, 
  "insulin_action_curve": 4, 
  "block_enable": false, 
  "selected_pattern": 0, 
  "auto_off_duration_hrs": 0, 
  "audio_bolus_size": 1.0, 
  "timeformat": 0, 
  "paradigm_enabled": 0, 
  "maxBolus": 16.0, 
  "patterns_enabled": false
Jason Calabrese
@jasoncalabrese
Sep 16 2016 01:36
try adding a meter, just use a fake number
@garykidd ^
garykidd
@garykidd
Sep 16 2016 01:37
yeah...did that.
garykidd
@garykidd
Sep 16 2016 01:44
so, again, i was able to hit some random buttons during mmtune and then I can get....
[
      "916.732", 
      2, 
      -69
    ],
Dana Lewis
@danamlewis
Sep 16 2016 01:45
A post on the new explorer board rig ICYMI: https://diyps.org/2016/09/15/openaps-rigs-are-shrinking-in-size/
Tim Mecklem
@tmecklem
Sep 16 2016 01:47
That looks great!
Is that using a single chip that does BLE and the sub GHz for the pump?
David Cintron
@loudestnoise
Sep 16 2016 01:50
@danamlewis Yay explorer board! 🦄🦄🦄🦄🦄🦄
@tmecklem AFAIK it's an Edison rig that just has the radio built in, negating need for TI stick
bummer though still won't work w/ Dex G5, unless you pull BGs from NS
garykidd
@garykidd
Sep 16 2016 01:55
anyone maybe know why my NS COB is drastically overstated..though a mouse-over indicates last carb intake, accurately? During a refresh of site, you can momentarily see COB displayed correctly--but then it jumps?
Tim Mecklem
@tmecklem
Sep 16 2016 02:00
Wait, does Edison do BLE?
I was trying to figure out how the openxshareble part could work.
(shows how little I know about the Edison I guess)
Has anyone investigated using an app on the phone to communicate NS glucose via bluetooth to the openaps rig instead of having to enable a hotspot?
Seems like a bit of overlap with the LoopKit stuff maybe.
Matt Walker
@walker0
Sep 16 2016 02:07
@tmecklem my edison runs over a bluetooth personal hotspot, it uses significantly less phone battery than the wifi personal hotspot
Tim Mecklem
@tmecklem
Sep 16 2016 02:07
I was thinking more like a serial BLE connection and not a general hotspot.
kudos on getting that working, btw! :)
Was just thinking that the Dex communicates with the phone, so it might be able to get the glucose numbers to the openaps rig directly without NS even if there’s no service around.
Scott Leibrand
@scottleibrand
Sep 16 2016 02:11
@tmecklem yeah, @loudestnoise is correct: it's an Edison base block with a built-in cc1110 to do the 900MHz comms. about the same size as a RL, but running a full rig, not just a BLE to 900MHz bridge
@garykidd might want to check your nightscout /api/v1/treatments.json
Tim Mecklem
@tmecklem
Sep 16 2016 02:15
That’s awesome. That looks like where I’m trying to head with these crazy SRF chips I’ve got :)
Scott Leibrand
@scottleibrand
Sep 16 2016 02:17
yeah, if you want a tiny linux box that can do 900MHz comms, you can't go much smaller than that. :)
David Cintron
@loudestnoise
Sep 16 2016 02:18
:+1:
Tim Mecklem
@tmecklem
Sep 16 2016 02:19
Is that a 3.7v lipo battery? Does the breakout have a charging circuit built in? So many questions :)
Chris Oattes
@cjo20
Sep 16 2016 02:19
Yes, and yes
Tim Mecklem
@tmecklem
Sep 16 2016 02:20
Does it power the Edison on 3.7 or does it convert the voltage to 5v?
Chris Oattes
@cjo20
Sep 16 2016 02:20
and the charger is higher current than the original Intel breakout board, so it is actually possible to charge a 2Ah battery in a sensible time
Edison runs on 3.2-4.5v
Tim Mecklem
@tmecklem
Sep 16 2016 02:20
oh, cool. I’m learning a lot.
Scott Leibrand
@scottleibrand
Sep 16 2016 02:21
it does not have a 5V level shifter, so it can't power USB OTG devices directly, but the next version will allow pass-through USB power from the UART port to the OTG port, so we can power USB OTG devices when a USB battery is plugged in there.
garykidd
@garykidd
Sep 16 2016 02:30
@scottleibrand @jasoncalabrese @bewest SO this is really interesting. Same issues with mmtune consistently delivery 99's except when buttons pushed during mmtune..and then I get a "good" freq. and it seems to work fine. Also, feel like it wasnt ti stick because I tried two..but also felt like it wasnt pump either because the same thing happened for a 712 pump in addition to the 715. So I had one other pump to try...my current pump...a 751. added that device, to my surprise it reported back "751" to my openaps use pump751 model and further...and more intersting...with no buttons pushed...openaps use pump751 mmtune worked just fine ...see below. So now I figure there is something about the pumps...is no one using these models...712, 715?
Here's the result of my openaps use pump751 mmtune
pi@raspberrypi:~/APS301 $ openaps use pump751 model
"751"pi@raspberrypi:~/APS301 $ opeuse pump751 mmtune
{
  "scanDetails": [
    [
      "916.300", 
      0, 
      -99
    ], 
    [
      "916.324", 
      0, 
      -99
    ], 
    [
      "916.348", 
      0, 
      -99
    ], 
    [
      "916.372", 
      0, 
      -99
    ], 
    [
      "916.396", 
      0, 
      -99
    ], 
    [
      "916.420", 
      0, 
      -99
    ], 
    [
      "916.444", 
      0, 
      -99
    ], 
    [
      "916.468", 
      4, 
      -50
    ], 
    [
      "916.492", 
      5, 
      -30
    ], 
    [
      "916.516", 
      5, 
      -28
    ], 
    [
      "916.540", 
      5, 
      -28
    ], 
    [
      "916.564", 
      5, 
      -28
    ], 
    [
      "916.588", 
      5, 
      -29
    ], 
    [
      "916.612", 
      5, 
      -32
    ], 
    [
      "916.636", 
      0, 
      -99
    ], 
    [
      "916.660", 
      0, 
      -99
    ], 
    [
      "916.684", 
      0, 
      -99
    ], 
    [
      "916.708", 
      0, 
      -99
    ], 
    [
      "916.732", 
      0, 
      -99
    ], 
    [
      "916.756", 
      0, 
      -99
    ], 
    [
      "916.780", 
      0, 
      -99
    ], 
    [
      "916.804", 
      0, 
      -99
    ], 
    [
      "916.828", 
      0, 
      -99
    ], 
    [
      "916.852", 
      0, 
      -99
    ], 
    [
      "916.876", 
      1, 
      -91
    ]
  ], 
  "setFreq": 916.564, 
  "usedDefault": false
Scott Leibrand
@scottleibrand
Sep 16 2016 02:33
@garykidd I suspect it's some other aspect of the TI stick radio settings that is poorly compatible with the particular older pumps you have. I've come across one that required a wider receive range, for example. seems to be a fair bit of manufacturing variability in the radios on these pumps
I used the ruby bin/mmtune to play with those settings, but ended up just using the pump that worked better without anything custom
garykidd
@garykidd
Sep 16 2016 02:37
but the 751 doesnt work with the openaps overall--right?
Scott Leibrand
@scottleibrand
Sep 16 2016 02:41
does it have a PC Connect under Utilities?
if so, you won't be able to set temps (or loop with it)
garykidd
@garykidd
Sep 16 2016 02:44
I assumed that's means not just directly under Utilities....cuase, yes, unfortunately, it is under Utilities,,,,Connect Devices,,,,,PC Connect
Kinda strange that I am the only one that has this issue...I guess I'm the only knuckle head using such old pumps...I think MM compelled me to trade my 723 in...sounds like that've worked just fine.
garykidd
@garykidd
Sep 16 2016 02:55
@scottleibrand would you be able to help me a little to get ruby/mmtune set up so that I could try to tiinker with it.....
Scott Leibrand
@scottleibrand
Sep 16 2016 02:59
there are others using old pumps. I don't think it's a model-specific thing
Scott Leibrand
@scottleibrand
Sep 16 2016 03:04
let's move this back to mmeowlink
garykidd
@garykidd
Sep 16 2016 03:08
sure...sorry
Scott Leibrand
@scottleibrand
Sep 16 2016 03:09
no worries, just trying to keep the context together. :)
Andrew Warrington
@andrew-warrington
Sep 16 2016 08:10
Hi all, some Medtronic clinical trial results for 670G: http://jama.jamanetwork.com/article.aspx?articleid=2552454
I think based on my own experience and what I've read from others in this group, that OpenAPS frequently performs as well or better. :) Hard to tell though as we don't collect data on e.g. hypos. Most people only talk about A1C and time in range.
Tim Street
@tim2000s
Sep 16 2016 10:26
I've just written up a co.pariaon of the Medtronic and OpenAPS publications.
Andrew Warrington
@andrew-warrington
Sep 16 2016 11:06
Great little article. Thanks. Totally agree that those of us here with poorer results may be self-selecting out of reporting. Also that Medtronic's algorithm has reason to be less aggressive - not just due
to liability, but also due to needing to manage an average population which is potentially less engaged than OpenAPS members.
Tim Street
@tim2000s
Sep 16 2016 11:10
Based on what I've seen on using SmartGuard, I think the latter has less part to play. It's very, very benign in regard to allowing recovery, erring massively on the side of caution. Given the US centricity of the new model, any risk of low and resulting damage that might cause a lawsuit will be being heavily avoided.
timomer
@timomer
Sep 16 2016 11:20
Hey all, I'm giving a talk about #wearenotwaiting next week in London at Internet of healthcare. On my final slide I want to show images of different users with their loop systems (closed/open). This is to show the many users and different systems out there and visually I think will be powerful. If you are happy to share such an image and for it to be publicly shown you can PM it to me. I would be very grateful!
I have very patchy internet right now so don't worry if I do not respond right away
Eric Jensen
@elnjensen
Sep 16 2016 12:41
Nice article, @tim2000s - thanks for sharing it.
Dana Lewis
@danamlewis
Sep 16 2016 13:22
Great post @tim2000s (although we didn't bring a product to market , as described at the end - very different), agree with a lot of your comparisons and thoughts on the limitations of available data
@timomer I can send you a dozen photos that people shared before for use at QSPH!
Mike
@libxmike
Sep 16 2016 13:23
@danamlewis is there any statistics how many use active open/close loop with openaps?
Dana Lewis
@danamlewis
Sep 16 2016 13:26
The only thing we track is people who voluntarily tell us when they've closed the loop, so the (n=1)*108+ number of DIY closed loopers is closed loop usage. I don't have any knowledge of number people in various forms doing open looping. And there's also people who close the loop but don't tell us, so that number is also a conservative low estimate
Mike
@libxmike
Sep 16 2016 13:30
@danamlewis yeah, its hart to keep tracking w/o some code in the app. About the Explorer Board, is it available somewhere to buy?
I'm building right now on Pi Zero and CareLink
but explorer board will be still smaller and extended range
Dana Lewis
@danamlewis
Sep 16 2016 13:31
And there's not even an app, since everyone's building themselves.Not yet but hopefully soon: https://twitter.com/danamlewis/status/776596835498545152
Tim Street
@tim2000s
Sep 16 2016 13:32
Hi @danamlewis - updated to read platform rather than product.
Dana Lewis
@danamlewis
Sep 16 2016 13:35
Thx @tim2000s!
Tim Street
@tim2000s
Sep 16 2016 14:13
If I could like your new blog post in Gitter, I would.
Dana Lewis
@danamlewis
Sep 16 2016 14:14
:)
Chris Oattes
@cjo20
Sep 16 2016 14:27
Just chuck a few :+1: around :)
Tim Street
@tim2000s
Sep 16 2016 14:28
👍👍👍
Tim Street
@tim2000s
Sep 16 2016 16:32
So here's a thought. Once I've got the glucose uploader functioning properly, I'll be getting my sgv data from the same source as SmartGuard uses. This means I'll be able to run the OpenAPS and SmartGuard algos side by side and have an exact comparison of how they intervene based on the same data!
Eric Jensen
@elnjensen
Sep 16 2016 17:04
Will the new Explorer board stack with other Sparkfun Edison blocks? I don't know how generic their interface is and whether it works with most things with that sort of connector (providing they don't try to use the same pins).
Dana Lewis
@danamlewis
Sep 16 2016 17:55
It's recorded in case you missed Ben's talk, but he'll be on a panel next if anyone wants to tune in!
Scott Leibrand
@scottleibrand
Sep 16 2016 18:04
@elnjensen yes, you can stack other sparkfun blocks in between the Edison and the Explorer board. It only has a bus connector on one side, though: same as the sparkfun battery block.
David
@davidkeddydb
Sep 16 2016 18:07
@danamlewis I can't believe how small you got it! Congratulations. Hopefully I will get there one day after I get my loop closed.
Scott Leibrand
@scottleibrand
Sep 16 2016 18:07
@tim2000s looking forward to that comparison! Are you running oref0 dev? When you go to do the comparison, you'll want to pay attention to whether autosens is detecting any sensitivity/resistance. Maybe also do comparisons both with and without meal assist enabled.
Hristina Dimova
@xpucuto
Sep 16 2016 18:51
@danamlewis @scottleibrand Last ngiht i had a low , i noticed something strange. The IOB on NS was -0.67 but the active insulin onthe pump was 0.5 . Usually i count on the IOB on NS website because it calculates the basal as well and i can decice how much carbs to take .
IMG_8275.jpg
IMG_8274.jpg
is it normal ?
Jason Wittmer
@jbwittmer
Sep 16 2016 19:13
@scottleibrand Trying to use your new script. Have openAPS installed but haven't done anything else. Getting error bash: cd: /home/pi/src: No such file or directory
any ideas?
Jason Wittmer
@jbwittmer
Sep 16 2016 19:23
I just did a mkdir ~/src and then it seemed to work
Scott Leibrand
@scottleibrand
Sep 16 2016 19:37
I thought I fixed that already. Did you pull latest oref0 dev first?
Jason Wittmer
@jbwittmer
Sep 16 2016 19:38
It said we had the latest version. For some reason we hadn't installed the json so it kicked us out. We installed json and now are trying to rerun script and getting an error
Scott Leibrand
@scottleibrand
Sep 16 2016 19:38
@xpucuto the pump IOB is only based on boluses. The net IOB repeated from OpenAPS to NS includes effect of temp basals. So that would be what you'd expect if you had lots of low-temps.
Jason Wittmer
@jbwittmer
Sep 16 2016 19:39
something to the effect of ImportError: No module names mmeowlink.vendors.mmeowlink then it saves Reinitialized existing Git repository in . . . . . and then On Branch Master and then Untracked files: ns.ini and then nothing added to commit but untracked files present Can't init /home/pi/myopenaps
Scott Leibrand
@scottleibrand
Sep 16 2016 19:41
@xpucuto also looks like you have a 0 basal profile configured in NS for some reason.
R. Dennis Steed
@rdsteed
Sep 16 2016 19:41

I find this community fascinating, but I find it hard to understand the jargon.I'm an endocrinologist and I've been caring for people with type 1 diabetes using home glucose monitoring, insulin pumps, and dosing algorithms for over 30 years. Many of the terms I thought I understood seem to mean something different here.

For example, I've always understood "IOB" to be an acronym for Insulin On Board, and is a quantity which is inherently greater than or equal to zero. Apparently it can mean something else here if it can be a negative number.

“When I use a word,” Humpty Dumpty said, in rather a scornful tone, “it means just what I choose it to mean—neither more nor less.” “The question is,” said Alice, “whether you can make words mean so many different things.” “The question is,” said Humpty Dumpty, “which is to be master—that’s all.” - Lewis Carroll

Scott Leibrand
@scottleibrand
Sep 16 2016 19:42
@rdsteed IOB on a pump does not include the effect of basal insulin.
So we chose to continue that, by introducing "net IOB", with 0 being the baseline of what your IOB would be with normal basal delivery (to match what the pump says)
AFAIK nobody considers "gross IOB", which would include IOB from basals. We could do that, but it would be even more confusing.
R. Dennis Steed
@rdsteed
Sep 16 2016 19:44
OK, but how could IOB ever be negative with any of those definitions?
Sandra Keßler
@SandraK82
Sep 16 2016 19:45
With a basal less then 100%
Jason Wittmer
@jbwittmer
Sep 16 2016 19:46
@scottleibrand Here's the error we're getting when trying to run the script again after installing json

Continue? y/[N] Y
Checking /home/pi/myopenaps: Traceback (most recent call last):
File "/usr/local/bin/openaps-use", line 4, in <module>
import('pkg_resources').run_script('openaps==0.1.5', 'openaps-use')
File "build/bdist.linux-armv7l/egg/pkg_resources/init.py", line 744, in run_script
File "build/bdist.linux-armv7l/egg/pkg_resources/init.py", line 1499, in run_script
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/EGG-INFO/scripts/openaps-use", line 63, 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-use", line 31, in configure_parser
available = devices.get_device_map(self.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 77, in FromConfig
vendor = vendors.lookup(config.get(candidate, 'vendor').split('.').pop( ), config)
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/vendors/init.py", line 27, in lookup
return get_map(config)[name]
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/vendors/init.py", line 22, in get_map
vendors = all_vendors(config)
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/vendors/init.py", line 36, in all_vendors
return get_vendors( ) + find_plugins(config)
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/vendors/init.py", line 16, in find_plugins
return [ v.get_module( ) for v in vendors ]
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/vendors/plugins/vendor.py", line 19, in get_module
return importlib.import_module(self.name)
File "/usr/lib/python2.7/importlib/init.py", line 37, in import_module
import(name)
ImportError: No module named mmeowlink.vendors.mmeowlink
Reinitialized existing Git repository in /home/pi/myopenaps/.git/
On branch master
Untracked files:
ns.ini

nothing added to commit but untracked files present
Can't init /home/pi/myopenaps

R. Dennis Steed
@rdsteed
Sep 16 2016 19:48
OK so IOB as used here means (the total insulin on board from my usual basals) minus (the actual insulin on board from the basals and boluses I have actually taken)?
Scott Leibrand
@scottleibrand
Sep 16 2016 19:48
@jbwittmer doesn't look like you have mmeowlink installed. the script is supposed to do that for you. can you do cd ~/src/oref0 && git pull and see if it pulls a new version
@rdsteed other way around. actual IOB from basals and boluses minus IOB from scheduled basals
Jason Wittmer
@jbwittmer
Sep 16 2016 19:49
It says Already up-to-date
@scottleibrand Sorry, says Already up-to-date
Scott Leibrand
@scottleibrand
Sep 16 2016 19:49
k, lemme check the code
@jbwittmer looks like you must be working on a directory that already exists, and has an openaps config referencing mmeowlink. try deleting the /home/pi/myopenaps/ directory and trying again.
Sandra Keßler
@SandraK82
Sep 16 2016 19:52
@rdsteed iob means only Bolus. Iob of Zero is just basal at 100%. Iob of -1 is one Unit below basal
Scott Leibrand
@scottleibrand
Sep 16 2016 19:52
@jbwittmer so it looks like a openaps use -h in /home/pi/myopenaps/ is failing, even before we do anything
Hristina Dimova
@xpucuto
Sep 16 2016 19:53
@scottleibrand I haven't update my basal rates to NS as I change them very often and the OpenAPS pulls data from the pumps and the receiver . It doesn't get data from the website . Do you think this could be the problem ? When i started looping in May the iob was displaying till 0 IOB . And if I had 0.4 and my basal was 0.8 I knew I wouldn't go low . I think I never had negative value . Probably With the recent update that changed I'm am not sure . But now When I have low I am afraid because I have no idea how much insulin I have left . I think the bolus shouldn't be subtracted if there is a zero temp . Bolus and temp basal act differently by my opinion . Sometimes I have noticed that if I had zero temp and put 4 UI it automatically says I have about 3 u IOB . And if you are right then why when there is negative iob my BG keep going down ? :)
Scott Leibrand
@scottleibrand
Sep 16 2016 19:54
@xpucuto it sounds like you got yourself into a "gross IOB" mindset by putting a zero basal profile into NS. with one of the recent NS updates, it switched from calculating IOB itself (based on your zero basal profile in this case) to using what OpenAPS provides, which is "net IOB"...
the thing to realize about negative IOB is it's not the same as negative insulin activity: it doesn't mean you should be going up yet. it just means that, over the course of the next 3-4h (DIA) you should eventually rise from having received less than your scheduled basal insulin
and of course, for that to be accurate, your scheduled basals on the pump have to at least approximately match your body's actual need for basal insulin.
R. Dennis Steed
@rdsteed
Sep 16 2016 19:56
@SandraK82 @scottleibrand I seem to be getting a slightly different messages here.
Jason Wittmer
@jbwittmer
Sep 16 2016 19:56
@scottleibrand OK we did that and restarted script and got this:
importing device file
Traceback (most recent call last):
File "/usr/local/bin/openaps-import", line 4, in <module>
import('pkg_resources').run_script('openaps==0.1.5', 'openaps-import')
File "build/bdist.linux-armv7l/egg/pkg_resources/init.py", line 744, in run_script
File "build/bdist.linux-armv7l/egg/pkg_resources/init.py", line 1499, in run_script
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/EGG-INFO/scripts/openaps-import", line 89, 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-import", line 81, in run
incoming = self.Make(candidate)
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/EGG-INFO/scripts/openaps-import", line 65, in Make
inst = Configurable.FromImport(configurable, self.config)
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/devices/device.py", line 66, in FromImport
vendor = vendors.lookup_dotted(fields['vendor'], config)
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/vendors/init.py", line 30, in lookup_dotted
vendors = all_vendors(config)
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/vendors/init.py", line 36, in all_vendors
return get_vendors( ) + find_plugins(config)
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/vendors/init.py", line 16, in find_plugins
return [ v.get_module( ) for v in vendors ]
File "/usr/local/lib/python2.7/dist-packages/openaps-0.1.5-py2.7.egg/openaps/vendors/plugins/vendor.py", line 19, in get_module
return importlib.import_module(self.name)
File "/usr/lib/python2.7/importlib/init.py", line 37, in import_module
import(name)
ImportError: No module named mmeowlink.vendors.mmeowlink
Could not import device.json
Scott Leibrand
@scottleibrand
Sep 16 2016 19:57
@jbwittmer what did you get after "Checking mmeowlink installation" ?
Jason Wittmer
@jbwittmer
Sep 16 2016 19:57
it said already up to date
hold on let me check that
Scott Leibrand
@scottleibrand
Sep 16 2016 19:58
you sure that was immediately after "Checking mmeowlink installation"? maybe paste that part
Jason Wittmer
@jbwittmer
Sep 16 2016 19:58
no sorry here it is
Checking mmeowlink installation
No module named mmeowlink.vendors.mmeowlink
mmeowlink.vendors.mmeowlink doesn't seem to be an importable python module
If it is a python module, try using --path to influence
PYTHONPATH
Hristina Dimova
@xpucuto
Sep 16 2016 19:58
@scottleibrand thanks for the explanation :) I have lowered my basal profile a lot since I started using OpenAPS and my TDD is much lower as well . Maybe I need test more carefully my basals and see how it turns out .
Scott Leibrand
@scottleibrand
Sep 16 2016 19:59
@xpucuto if you notice OpenAPS usually giving you low-temps at a certain time of day, that's a good indication that the basals 60-90m previous to that timeframe might be too high. (and vice versa, if carbs aren't involved)
@jbwittmer ok, sounds like the automatic mmeowlink install is failing. can you try a few things for me to debug why?
Jason Wittmer
@jbwittmer
Sep 16 2016 20:00
sure
Scott Leibrand
@scottleibrand
Sep 16 2016 20:01
openaps vendor add --path . mmeowlink.vendors.mmeowlink || (echo Install mmeowlink)
Hristina Dimova
@xpucuto
Sep 16 2016 20:01
@scottleibrand :+1: I will have that in mind but usually the problem is after I enter wrong calculated bolus .
Jason Wittmer
@jbwittmer
Sep 16 2016 20:01
@scottleibrand here's what I get pi@raspberrypi:~/myopenaps $ openaps vendor add --path . mmeowlink.vendors.mmeowlink || (echo Install mmeowlink)
No module named mmeowlink.vendors.mmeowlink
mmeowlink.vendors.mmeowlink doesn't seem to be an importable python module
If it is a python module, try using --path to influence
PYTHONPATH
Scott Leibrand
@scottleibrand
Sep 16 2016 20:02
ok, so it never said "Install mmeowlink", which means that command isn't returning the error code I expected it to
one sec
try openaps vendor add --path . mmeowlink.vendors.mmeowlink 2>&1 | grep "No module" && echo Install mmeowlink
Jason Wittmer
@jbwittmer
Sep 16 2016 20:04
@scottleibrand I tried running from root and from myopenaps and got the same message each time:
pi@raspberrypi:~ $ openaps vendor add --path . mmeowlink.vendors.mmeowlink 2>&1 | grep "No module" && echo Install mmeowlink
No module named mmeowlink.vendors.mmeowlink
Install mmeowlink
pi@raspberrypi:~ $ cd myopenaps
pi@raspberrypi:~/myopenaps $ openaps vendor add --path . mmeowlink.vendors.mmeowlink 2>&1 | grep "No module" && echo Install mmeowlink
No module named mmeowlink.vendors.mmeowlink
Install mmeowlink
pi@raspberrypi:~/myopenaps $
Scott Leibrand
@scottleibrand
Sep 16 2016 20:04
ok, good
let me go update the script to do it that way
ok, updated. please do another git pull in ~/src/oref0 and try again
Jason Wittmer
@jbwittmer
Sep 16 2016 20:06
@scottleibrand should I delete the directory again before I start?
Scott Leibrand
@scottleibrand
Sep 16 2016 20:06
sure
Jason Wittmer
@jbwittmer
Sep 16 2016 20:07
and then this:
cd ~/src && git clone -b dev git://github.com/openaps/oref0.git || (cd oref0 && git checkout dev && git pull)
?
Scott Leibrand
@scottleibrand
Sep 16 2016 20:07
sure
@xpucuto if you give too large of a bolus, then once oref0 sees you starting to drop, it should low-temp aggressively to counteract that. eventually that will bring you back up to target, but it will take hours
Jason Wittmer
@jbwittmer
Sep 16 2016 20:08
OK it updated something
@scottleibrand here's what I got. Go ahead with the second part of the script now?
pi@raspberrypi:~ $ cd ~/src && git clone -b dev git://github.com/openaps/oref0.git || (cd oref0 && git checkout dev && git pull)
fatal: destination path 'oref0' already exists and is not an empty directory.
Already on 'dev'
Your branch is up-to-date with 'origin/dev'.
remote: Counting objects: 11, done.
remote: Compressing objects: 100% (11/11), done.
remote: Total 11 (delta 2), reused 0 (delta 0), pack-reused 0
Unpacking objects: 100% (11/11), done.
From git://github.com/openaps/oref0
1775426..ebae814 dev -> origin/dev
Updating 1775426..ebae814
Fast-forward
bin/oref0-setup.sh | 2 +-
lib/profile/carbs.js | 5 +++++
2 files changed, 6 insertions(+), 1 deletion(-)
Scott Leibrand
@scottleibrand
Sep 16 2016 20:09
yep
Jason Wittmer
@jbwittmer
Sep 16 2016 20:09
and run it from the root directory, correct?
Scott Leibrand
@scottleibrand
Sep 16 2016 20:10
yeah, that's easiest
C-Ville
@C-Ville
Sep 16 2016 20:11
@christerjensen sorry for responding late, have been on vacation. It is a 715 from Poland as far as I could retrieve the heritage of the pump, but it uses mg/dL so no conversion issues for me there, sorry
Jason Wittmer
@jbwittmer
Sep 16 2016 20:13
@scottleibrand Seems to have workeed. Thanks. I'll update you in a bit
Scott Leibrand
@scottleibrand
Sep 16 2016 20:13
@rdsteed so this is the easiest way IMO to understand net IOB: if you normally have a basal of 1.0 U/hr, and OpenAPS low-temps you to 0 U/hr for 30m, then your net IOB immediately afterward is about -0.5U (actually like -0.48U due to IOB decay).
R. Dennis Steed
@rdsteed
Sep 16 2016 20:17
@scottleibrand Thanks, I understand the nomenclature now. I'm troubled, though, by the concept of using an arbitrary "normal basal" as the basis for net IOB.
Scott Leibrand
@scottleibrand
Sep 16 2016 20:19
why do you think that the normally scheduled basal is arbitrary?
FWIW, we do have an automatic sensitivity detection feature that adjusts the normal basal for purposes of setting temp basals. however, to avoid feedback effects, we do not use that adjustment directly in calculating net IOB. Instead, we adjust the decision-making after IOB calculation, both by using the adjusted basal as the baseline from which to calculate further temps, and also (if enabled) to adjust the target BG.
R. Dennis Steed
@rdsteed
Sep 16 2016 20:21
I consider normally scheduled basal to be an arbitrary decision of the person who programmed the pump.
Dana Lewis
@danamlewis
Sep 16 2016 20:22
all of the variables are arbitrary decisions of people - ISF, carb ratios, basal rates, etc.
Scott Leibrand
@scottleibrand
Sep 16 2016 20:22
so you'd like the APS to automatically learn what basals and ratios should be? we've written up an approach for how to do that, but haven't implemented it yet: openaps/oref0#99
I'm not sure we'll ever want to give the APS full autonomy to adjust ratios without review by a human, but that approach should at least allow it to make non-arbitrary recommendations
but even without that, remember that we're trying to do better than a dumb pump that's programmed to deliver a certain basal rate, regardless of BG etc. By making all of our adjustments relative to that pre-programmed basal schedule, we're able to deliver significantly better outcomes than if you turn off the OpenAPS adjustments and simply let the pump deliver its normal basals
R. Dennis Steed
@rdsteed
Sep 16 2016 20:27
I probably should have said "potentially arbitrary". The net IOB represents the difference between the insulin on board delivered by the closed loop and the insulin on board that would be present if the pump was running open loop.
Scott Leibrand
@scottleibrand
Sep 16 2016 20:29
and if no basals were given, yes
would love your input on how to better deal with the info we have to make better decisions or present information to the user in an easier-to-understand way
R. Dennis Steed
@rdsteed
Sep 16 2016 20:30
Would love to give input. But first we have to speak the same language ;-)
Dana Lewis
@danamlewis
Sep 16 2016 20:31
@rdsteed would also appreciate your take, after reading the reference design, on how to make this more clear
R. Dennis Steed
@rdsteed
Sep 16 2016 20:34
I've looked over the reference design, and I've read a lot of the code. My first thought is that I would like to have a better separation of the underlying algorithms and the computer programs.
Scott Leibrand
@scottleibrand
Sep 16 2016 20:37
@rdsteed are you aware of any existing language we could be using that describes these concepts? We weren't aware of anyone already talking about such things when we started DIYPS and OpenAPS, so we introduced some descriptive terms like "net IOB", and some obvious ones like "COB", to describe the concepts we felt were necessary for average PWDs to understand the effects of an APS (rather than just treating it like a black box only comprehensible to engineers, which is what most researchers and companies seem to be doing). But if anyone actually is talking about such things, we'd be more than happy to try to use the existing language where appropriate.
R. Dennis Steed
@rdsteed
Sep 16 2016 20:43
The endocrine community has talked about these things for years, and we don't always agree on the language. I recently read a paper describing three different definitions of "insulin sensitivity " - and I didn't completely agree with any of them. As endocrinologists, we struggle with the different ways the pump manufacturers have implemented their insulin on board calculators and correction doses. It's just an ongoing problem. As a newb to the DIY online community, I'm just trying to translate into the terms I'm used to.
Scott Leibrand
@scottleibrand
Sep 16 2016 20:48
fair enough. for us, "insulin sensitivity" is just ISF: the number of mg/dL BG is expected to fall (over ~4h) for each unit of extra insulin delivered. what are the other definitions?
R. Dennis Steed
@rdsteed
Sep 16 2016 20:51
I'll try to get you a reference to the paper, don't have it at my fingertips. But I prefer an insulin sensitivity definition that is independent of a person's size. The ISF for people with normal physiology can vary widely based on a person's size.
Scott Leibrand
@scottleibrand
Sep 16 2016 20:52
so instead of mg/dL per U, you want to see mg/dL per U per kg?
R. Dennis Steed
@rdsteed
Sep 16 2016 20:54
No, I actually would prefer it to be grams/unit (which is, oddly enough, the carb ratio). People with typical insulin sensitivity will have a carb ratio someplace around 10 or so independent of their size.
Scott Leibrand
@scottleibrand
Sep 16 2016 20:54
grams of what per unit?
R. Dennis Steed
@rdsteed
Sep 16 2016 20:55
grams of carbohydrate per unit of insulin - the carb ratio.
Scott Leibrand
@scottleibrand
Sep 16 2016 20:55
sure, but how do you do that for ISF?
what would make the most sense to me would be to go the other way, and look at a Carb Sensitivity Factor (CSF): the number of mg/dL of BG rise expected for each g of carbohydrates
that way ISF and CSF are directly comparable, and the IC ratio is the ratio of ISF to CSF
because when sensitivity changes in the body due to exercise or hormones or whatever, we usually see ISF change, but CSF stay relatively constant (so IC changes as well)
right now we're having to try to adjust ISF and IC together, but if we used the underlying CSF instead of the IC, we'd only have to adjust one thing
R. Dennis Steed
@rdsteed
Sep 16 2016 20:57
Yes, that does make sense for the individual. But when we speak of the concept of being insulin insensitive (insulin resistant), it is nice to have a definition that works for people large or small.
You can think of the ISF (I usually use ICF for insulin correction factor) and the Carb Sensitivity factor as both being linked by the body dimensions (ie fairly constant numbers) to an underlying single insulin sensitivity.
Scott Leibrand
@scottleibrand
Sep 16 2016 21:00
yeah, that makes sense. let me think through the units
R. Dennis Steed
@rdsteed
Sep 16 2016 21:01
OK. Think about it. Be glad to discuss further.
Scott Leibrand
@scottleibrand
Sep 16 2016 21:02
so, clinically, an ISF of 40 mg/dL for someone weighing 75 kg would be equivalent sensitivity to an ISF of 75 for someone weighing 40 kg?
R. Dennis Steed
@rdsteed
Sep 16 2016 21:05
More or less. Probably ab pretty good rule of thumb
Scott Leibrand
@scottleibrand
Sep 16 2016 21:06
the units on CSF would be mg/dL per g, so the mg of BG and g of carbs would cancel out, and CSF units would just be 1/volume. which kinda makes sense, if you think of the body as a big bag of liquid that you're dissolving glucose into
R. Dennis Steed
@rdsteed
Sep 16 2016 21:06
Another way of thinking about it is that the insulin concentration in the blood is simply a reflection of the the grams of glucose dissolved in the body's water. Big people have more water that small people.
Scott Leibrand
@scottleibrand
Sep 16 2016 21:06
so I think CSF already accomplishes your goal of being body size independent?
heh, jinx
R. Dennis Steed
@rdsteed
Sep 16 2016 21:07
Exactly.
jinx indeed.
Scott Leibrand
@scottleibrand
Sep 16 2016 21:07
really glad to hear that wasn't an original idea of mine. :)
R. Dennis Steed
@rdsteed
Sep 16 2016 21:10
Yes, a lot of these ideas have been kicking around for years. The cool thing is that the people in the DIY community are doing it. Also, the sensors have sucked until recently. I explored doing closed loop with the Glucowatch, but that thing was scary inaccurate.
Scott Leibrand
@scottleibrand
Sep 16 2016 21:10
so if we do a similar exercise with ISF, the mg/dL per U actually has to be multiplied by body weight, not divided (if you want 40mg/dL/U * 75kg = 75kg * 40mg/dL/U) so you get kg*mg/dL/U, right?
Peter Miller
@ochenmiller
Sep 16 2016 21:10
@rdsteed , could you explain the somewhat "magic ratio" of 3:1 for ISF and CSF? Does the fact that ISF is somewhat independent of body mass play into that?
R. Dennis Steed
@rdsteed
Sep 16 2016 21:12
I think that the magic ratio is just a reflection of the fact that most people are fairly similar in size and responsiveness to insulin.
Scott Leibrand
@scottleibrand
Sep 16 2016 21:12
@ochenmiller 3:1 ISF:CSF? or 3:1 ISF:IC? I'm not familiar with this "magic ratio" myself
Peter Miller
@ochenmiller
Sep 16 2016 21:13
Our endo floated that concept the first time we met him. As a baseline for determining our numbers.
And the empirical evidence has shown that to be pretty accurate - for us at least
We've also used it in the reverse. eat 10 carbs, your BG goes up 30 points. ish.
Scott Leibrand
@scottleibrand
Sep 16 2016 21:15
@ochenmiller ok, so you're talking about ISF:IC ratio, which is the CSF I was talking about.
Peter Miller
@ochenmiller
Sep 16 2016 21:16
right
Scott Leibrand
@scottleibrand
Sep 16 2016 21:16
for @danamlewis it's 4 rather than 3, but agreed, something in that range seems about right
Peter Miller
@ochenmiller
Sep 16 2016 21:17
I find it interesting that it's a fairly predictable and linear ratio.
R. Dennis Steed
@rdsteed
Sep 16 2016 21:17
OK. Then IC is what I consider to be "insulin sensitivity".
Sandra Keßler
@SandraK82
Sep 16 2016 21:18
Hm interesting, I need 3-4 units for 14 carbs (Germans learned one BE (bread unit)). Ergo one BE let me rise around 100
(One bread unit equals roughly 14g carbs)
R. Dennis Steed
@rdsteed
Sep 16 2016 21:22
@SandraK82 mmol vs mg / dl, BE vs g of cho - gets pretty involved.
Sandra Keßler
@SandraK82
Sep 16 2016 21:22
Uh sorry, that mixes a lot
R. Dennis Steed
@rdsteed
Sep 16 2016 21:23
@SandraK82 yes - head explodes!
Sandra Keßler
@SandraK82
Sep 16 2016 21:23
I'm just getting started with openaps and all of that. Before that I never cared about mmol and all that stuff
Mike
@libxmike
Sep 16 2016 21:24
I thing everyone should stay with mg/dl and 10g carbs as units, still don't understand why mmol is used
Sandra Keßler
@SandraK82
Sep 16 2016 21:25
Currently exploring my options. Got a cheap 522 and now looking forward to build a loop
R. Dennis Steed
@rdsteed
Sep 16 2016 21:25
Same reason they use meters and we use feet.
Sandra Keßler
@SandraK82
Sep 16 2016 21:26
Ultimately I want to loop with libre and Roche combo
Mike
@libxmike
Sep 16 2016 21:26
in engineering/science /physic etc we used metrics in US
Sandra Keßler
@SandraK82
Sep 16 2016 21:27
And of course iOS
R. Dennis Steed
@rdsteed
Sep 16 2016 21:28
I gather that loudnate is the guru of closing the loop with iOS.
Scott Leibrand
@scottleibrand
Sep 16 2016 21:30
problem with "just use metric" as a heuristic is that mmol/L and mg/dL are both metric, one is just "normal" units and one is "chemist" (US meaning of chemical scientist, not UK meaning of pharmacist) units
R. Dennis Steed
@rdsteed
Sep 16 2016 21:31
Yes, bad analogy. I should have said that the units of measure are largely traditional.
Scott Leibrand
@scottleibrand
Sep 16 2016 21:31
yep
R. Dennis Steed
@rdsteed
Sep 16 2016 21:32
And sometimes it is dangerous - like when people accidentally switch their meter mode from mg/dl to mmol/l.
Scott Leibrand
@scottleibrand
Sep 16 2016 21:32
even most of the metric system is based on an arbitrarily-sized kg
Sandra Keßler
@SandraK82
Sep 16 2016 21:32
Yeah loudnate is something. I have not yet decided
Need to crack the combo and/or build an ble transmitter for the 522
R. Dennis Steed
@rdsteed
Sep 16 2016 21:36
@SandraK82 was there a typo there in your comment about the 522?
Sandra Keßler
@SandraK82
Sep 16 2016 21:37
@rdsteed what do you mean
Scott Leibrand
@scottleibrand
Sep 16 2016 21:37
@SandraK82 have you tried the RileyLink?
Sandra Keßler
@SandraK82
Sep 16 2016 21:39
Not yet. Currently low on money because of the development of the blueReader
R. Dennis Steed
@rdsteed
Sep 16 2016 21:39
I wasn't sure what you meant by "combo" or "an ble". I'm guessing "an ble" refers to a bluetooth low energy link, but doesn't the 522 use a proprietary 900mHz link rather than bluetooth?
Sandra Keßler
@SandraK82
Sep 16 2016 21:40
Looking into the nrf9e5 at the moment
Yeah meaning an custom designed adapter for ble to 900 mhz
Scott Leibrand
@scottleibrand
Sep 16 2016 21:42
@rdsteed the Spirit Combo is another pump. the RileyLink (and things like the nrf9e5 or even the upcoming 915MHz Edison Explorer board) can be used to bridge 900MHz pump comms to BLE
Sandra Keßler
@SandraK82
Sep 16 2016 21:42
The 9e5 is very cheap and does not need a programmer
The combo pump/meter communicate via Bluetooth 2.0 edr. We just need to figure out the pairing
Scott Leibrand
@scottleibrand
Sep 16 2016 21:44
@SandraK82 looks like the nRF9E5 is subg-only, no BLE?
looks very similar to the TI cc1110
Sandra Keßler
@SandraK82
Sep 16 2016 21:46
Yes @scottleibrand however I'm very familiar with the nrf5x series of ble chips. So combine a 9e5 with the nrf51 results in very tiny 900mhz to ble bridge
Scott Leibrand
@scottleibrand
Sep 16 2016 21:46
:+1:
"very tiny" = smaller than RileyLink?
Sandra Keßler
@SandraK82
Sep 16 2016 21:47
The 9e5 is software defined. I do not need any programmer to use it
Did you see latest blueReader prototypes?
50% smaller then limitter
Scott Leibrand
@scottleibrand
Sep 16 2016 21:47
where does the software run? cc1110 basically puts the software on the chip via the debugger (or now via bit-bang protocol from an Edison)
no, I haven't been following bluereader or limttter development, as we can't get Libre over here
Sandra Keßler
@SandraK82
Sep 16 2016 21:48
It runs on the nrf m0 arm cortex
Scott Leibrand
@scottleibrand
Sep 16 2016 21:50
and how do you interface with that?
Sandra Keßler
@SandraK82
Sep 16 2016 21:51
Tomorrow I will post new images on http://unendlichkeit.net
The programmer is the nrf51-dk. The connection onboard is serial between chips as it is easier then spi
Spi would be another option currently used by the limitter
R. Dennis Steed
@rdsteed
Sep 16 2016 21:55
@SandraK82 blueReader looks very cool. I always thought the Abbott Navigator was the best sensor technology, though it was not very user friendly. How do you like the Libre?
Sandra Keßler
@SandraK82
Sep 16 2016 21:57
I use it now for over two years, paid by myself. Like it very much. Helps a lot to get better a1c ( down from 7,5-9 to <7 in the last 7 quarter years)
R. Dennis Steed
@rdsteed
Sep 16 2016 21:58
@SandraK82 Does how often you query the Libre affect its battery life?
Sandra Keßler
@SandraK82
Sep 16 2016 22:01
Now it does not. The libre measurement is covered by an internal battery. The read is powered by the reader.
Sorry have to go for now. It's Friday night in germany and I'm at a club 😹🎉
Scott Leibrand
@scottleibrand
Sep 16 2016 22:05
enjoy! :)
R. Dennis Steed
@rdsteed
Sep 16 2016 22:06
@SandraK82 OK. Thanks for the info. Was concerned that waking Libre frequently would deplete its battery sooner. Have a good one!
Sandra Keßler
@SandraK82
Sep 16 2016 22:22
Thanks a lot 😊
Jason Wittmer
@jbwittmer
Sep 16 2016 23:02
@scottleibrand OK so the script worked partway. Things we've figured out so far - the pump serial number wasn't correct in pump.ini - had to edit that manually
Also had to go in and manually correct the order of crrbhistory.json and basal profile.json in the meal.json and oref0-meal in the ns-loop
Now we can run a ns-loop and pull cgm data /etc from Nightscout and can run pump loop and get it to set a temp basal but when I run ns-loop again the upload is failing. Is there somewhere else I need to manually change NIGHTSCOUT)host and API_SECRET?
Scott Leibrand
@scottleibrand
Sep 16 2016 23:15
yes, for now you need to add those to the crontab
should be fairly easy for me to make the script do that too though
Jason Wittmer
@jbwittmer
Sep 16 2016 23:15
that's what I had figured and was working on - I'll let you know if it works
Scott Leibrand
@scottleibrand
Sep 16 2016 23:16
the one in the crontab is the hashed secret, not the plain text one
Jason Wittmer
@jbwittmer
Sep 16 2016 23:16
yep. just did it . i'll run the loop again and see if it updates
Jason Wittmer
@jbwittmer
Sep 16 2016 23:28
@scottleibrand Still not working. Hmm. Any other ideas? It pulls data from Nightscout fine. Just can't upload
Scott Leibrand
@scottleibrand
Sep 16 2016 23:29
ok, updated it to put the env vars in crontab. you can test if you want by doing cd ~/src/oref0 && git checkout oref0-setup && git pull and then re-running it
Jason Wittmer
@jbwittmer
Sep 16 2016 23:29
I've already put them there but still not uploading.
Scott Leibrand
@scottleibrand
Sep 16 2016 23:29
how did you generate the hash?
Jason Wittmer
@jbwittmer
Sep 16 2016 23:29
Path doesn't match the path on my rig - don't know if that makes a difference or not. They are both Raspberry pi 3
Scott Leibrand
@scottleibrand
Sep 16 2016 23:30
I don't think it matters, but I can fix that too
Jason Wittmer
@jbwittmer
Sep 16 2016 23:30
I did the method recommended in the docs echo - n "blahblah" | shasum
Scott Leibrand
@scottleibrand
Sep 16 2016 23:31
ok, that should be the same as nightscout hash-api-secret "blahblah", which is what I just added
Eric Jensen
@elnjensen
Sep 16 2016 23:32
@scottleibrand @rdsteed Just wanted to say that I really enjoyed reading through your conversation. I talk a lot about the importance of clear units in the astronomy classes I teach, and with my research students. I do think that sometimes there's a place for "user-friendly" units, even if they aren't the most dimensionally reduced or the most general. For example, you could reduce miles per gallon (length divided by volume) to an inverse area, but it's not that enlightening to do so. Similarly, I definitely think there's a place for the kind of insulin-sensitivity or carb-sensitivity units we use - you could convert them to be more general and independent of body size (and maybe that has a useful place in the code), but at the end of the day (or at dinner time!) I need to know how much insulin to take for the meal I'm about to eat, and I'm not as interested in how that would scale for someone of a different body size.
The trick (as always in clear communication) is to figure out who your audience is, and how to frame your message in a way that connects with them.
Not always easy to do, and not always possible to craft something that is most resonant/accessible for all audiences.
Scott Leibrand
@scottleibrand
Sep 16 2016 23:34
@jbwittmer I just updated to use the shell $PATH
Jason Wittmer
@jbwittmer
Sep 16 2016 23:35
@scottleibrand Not sure I understand. I did check 'env' and got the correct path and put that in the cron but still didn't work. Something else I need to do?
Scott Leibrand
@scottleibrand
Sep 16 2016 23:35
just telling you I fixed the script to use the same PATH as in env (edited to clarify)
Jason Wittmer
@jbwittmer
Sep 16 2016 23:36
when I use openaps use ns shell preflight it returns true. Still won't upload thought
Scott Leibrand
@scottleibrand
Sep 16 2016 23:36
my upload stuff doesn't (yet) use ns
Jason Wittmer
@jbwittmer
Sep 16 2016 23:37
yes I know, just thought that might help you know whether we had environmental variables set correctly and ? any other place to try to change things so it can upload.
Scott Leibrand
@scottleibrand
Sep 16 2016 23:39
try openaps upload-recent-treatments
Jason Wittmer
@jbwittmer
Sep 16 2016 23:40
it says entries.json does not exist - no recent treatments to upload. We're still running with pump disconnected at this point if that makes a difference
Scott Leibrand
@scottleibrand
Sep 16 2016 23:40
that means your NIGHTSCOUT_HOST is probably wrong
Jason Wittmer
@jbwittmer
Sep 16 2016 23:41
OK. I'll check it. But why can we download from the nightscout site in that case?
Scott Leibrand
@scottleibrand
Sep 16 2016 23:41
because that's using the ns device
I converted that part over, but haven't gotten to the rest of it yet. you're welcome to help if you like once you have it working. :)
Jason Wittmer
@jbwittmer
Sep 16 2016 23:42
just checked by copying directly from the crontab and pasting into browser. The HOST is correct in the crontab
@scottleibrand would there be any reason I couldn't switch out the new method for the old in your ns-loop alias? Would that have other ill effects you can think of?
Scott Leibrand
@scottleibrand
Sep 16 2016 23:44
it'll be a bit of work to get it all working, but well worth it if you succeed and can export the results for inclusion in the oref0-setup template
in other words, please do. :)
Jason Wittmer
@jbwittmer
Sep 16 2016 23:46
I'll try. Problem is I'm trying to help get this new family up and running and I'm not sure if I should keep hammering away at using your script or just revert to what I know at this point. And I need to get back home for my wife's birthday tomorrow (4 hours away). Guess I'll have to make a choice soon.
Scott Leibrand
@scottleibrand
Sep 16 2016 23:50
I think the work to get NS uploads working should be the same either way
so if you're comfortable running through that, you can probably knock it out pretty easily against my setup
Chris Oattes
@cjo20
Sep 16 2016 23:56
@elnjensen :+1: Most diabetics have a feel for how much 1u drops them and how much insulin they need to take to cover 1g carbs, and the effect of changing those is intuitive. Analysis of the relationship between those two numbers and bodyweight may help give some insight on how you might improve your ratios, but when you're going high after meals a simple "I'm going to increase the amount I do per g carb" is far easier (for most people) to make informed decisions about
Scott Leibrand
@scottleibrand
Sep 16 2016 23:57
yeah. the question for me becomes: how do you respond to changes in sensitivity over time? do you change ISF, IC ratio, basal schedule, or a combination of all three?
from a physiological perspective, it appears that a 20% increase in sensitivity should correspond to a 20% decrease in basal needs, a 20% increase in ISF, and a 20% decrease in IC ratio. I don't think most people change all three together, though.