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

6th
Nov 2015
Jason Calabrese
@jasoncalabrese
Nov 06 2015 00:13
@ochenmiller thanks, I'm not sure what the name of the clip style is on the inset 30, and no idea what the standards are there, but my goal is to use the same site and find something that clips into is and connects to the MM
Dana Lewis
@danamlewis
Nov 06 2015 00:13
wondering, @bewest @mddub or anyone else - is there a write up comparing MM Connect vs. using openaps tools to upload pump data into Nightscout? are they 1:1 on all data going up, or is there a difference in what we can pull off the pump from Connect?
Scott Leibrand
@scottleibrand
Nov 06 2015 00:14
MM Connect is more like using RileyLink to sniff the MySentry packets.
Dana Lewis
@danamlewis
Nov 06 2015 00:15
Yes, wanting an exact comparison list though ;) if there's one handy someone's already typed together.
Scott Leibrand
@scottleibrand
Nov 06 2015 00:16
That doesn't get you the full pumphistory, just the basic info beaconed every 5m. Not sure if @mmdub wrote up a user-friendly list when he did his web-scraper...
but it's basically BG and reservoir data
Mark Wilson
@mddub
Nov 06 2015 00:18
@danamlewis @scottleibrand so it depends on what data, at what level of real-time, and whether you mean with medtronic's app or figuring out how to talk to it ourselves
the default setup (using their app) sends mysentry status data every 5 minutes. there's a bit more than BG and reservoir, but not much https://github.com/mddub/minimed-connect-to-nightscout/blob/master/test/_samples.js
Dana Lewis
@danamlewis
Nov 06 2015 00:19
Keep getting asked by non-devs about how to get Medtronic pump info into NS. Would like a side by side of the data we can get in there, and basics on the different methods.
Wanted to ask first in case someone has already written something, vs. starting with a blank page.
Mark Wilson
@mddub
Nov 06 2015 00:19
the default setup (using their app) also triggers a history read once a day, which goes straight into carelink
okay, i could write this up in a more structured format if that's useful
many of the cases are "we could do this but no one has done it yet"
Dana Lewis
@danamlewis
Nov 06 2015 00:20
That would be awesome! (I was hoping someone who's spent more time in it would have done or be willing :))
Mark Wilson
@mddub
Nov 06 2015 00:22
but it would be based only on my experience with connect + NS; @bewest would know more about the role openaps should play
Dana Lewis
@danamlewis
Nov 06 2015 00:24
That would be a great starting place, though - so will take anything you're willing to write up :) and @bewest, me, others can add to it if needed
Ben West
@bewest
Nov 06 2015 00:27
mark likes using connect to get pump data into NS
this will work but, you have to trust Medtronic to give you data that Medtronic thinks you should get, but it's very similar to the share bridge
Dana Lewis
@danamlewis
Nov 06 2015 00:27
yep, but some people don't have or can't afford connect (hearing that more and more), so would be nice to show the options
Mark Wilson
@mddub
Nov 06 2015 00:28
yeah, there are also many speculative uses of a connect which would be better than going through carelink
Ben West
@bewest
Nov 06 2015 00:28
it doesn't use any equipment directly, it copies data for Medtronic API into NS
yeah this is just using the Medtronic as a product
as they provide it
I like the idea of using openaps as a Nightscout uploader
seeing pump unified with glucose in real time is going to help a lot of people
and it's 2015 already
Dana Lewis
@danamlewis
Nov 06 2015 00:29
+1
Ben West
@bewest
Nov 06 2015 00:29
the way to configure it to upload to NS right now.... the steps to upload to NS are fairly complicated and can be simplified
Mark Wilson
@mddub
Nov 06 2015 00:30
i imagine within a couple months there won't be a case where a connect is a better option than a rileylink
Ben West
@bewest
Nov 06 2015 00:30
and it'd be nice to have that be the accessible and safe introduction to openaps
Dana Lewis
@danamlewis
Nov 06 2015 00:30
+1
Ben West
@bewest
Nov 06 2015 00:30
monitor, predict, control :-)
Mark Wilson
@mddub
Nov 06 2015 00:30
+1
Ben West
@bewest
Nov 06 2015 00:31
I'm still using carelink stick
https://github.com/bewest/mmblelink/tree/wip/bewest/decocare this branch has a mmbelink-send tool that is the same, mostly as mm-send-comm
also has openaps vendor
still having bit of trouble with wakeup command, once it's on it seems to work great
it's slower than the carelinkstick
would be interesting to RE the connect gadget as welll
Mark Wilson
@mddub
Nov 06 2015 00:39
@bewest @danamlewis this is my contribution from the mmconnect side of things
Ben West
@bewest
Nov 06 2015 00:42
nice
looks good
also speculative
Mark Wilson
@mddub
Nov 06 2015 00:42
for the convenience of carrying around just a connect and being able to see bolus times and amounts (inferred based on iob) on my watchface in realtime, it's pretty nice
Ben West
@bewest
Nov 06 2015 00:42
finishing RE for contour nextlink
also, finishing patches in android-uploader to work with carelink stick, lol
but those are fairly speculative I guess
Mark Wilson
@mddub
Nov 06 2015 00:43
heh, less speculative than my speculative options
Dana Lewis
@danamlewis
Nov 06 2015 00:44
nice @mddub
Mark Wilson
@mddub
Nov 06 2015 00:44
i guess if you had regular mmcsv polling, then we've basically replaced tidepool
Ben West
@bewest
Nov 06 2015 00:44
oh, your connect plugin won't add teh history?
once a day?
Mark Wilson
@mddub
Nov 06 2015 00:45
nope
Scott Leibrand
@scottleibrand
Nov 06 2015 00:45
We asked them (Medtronic) for 0.1U resolution on reservoir info.
Ben West
@bewest
Nov 06 2015 00:45
oh I see
Scott Leibrand
@scottleibrand
Nov 06 2015 00:45
We'll see what they say at D-Data.
Mark Wilson
@mddub
Nov 06 2015 00:45
the main priority was sgv
Ben West
@bewest
Nov 06 2015 00:45
you only hit their json endpoint
gotcha
is that even in the json endpoint?
Mark Wilson
@mddub
Nov 06 2015 00:45
as you've seen, it's been complicated enough to show pump iob from the json in the UI
Ben West
@bewest
Nov 06 2015 00:46
hehe, yeah
Mark Wilson
@mddub
Nov 06 2015 00:46
there wasn't any ability to visualize basals at the time, either
now that we have that, it's more appealing, but i still don't want to write an adapter
Ben West
@bewest
Nov 06 2015 00:46
hmmm
maybe we can require('mmcsv') or something
or put it in dcalc
Mark Wilson
@mddub
Nov 06 2015 00:47
yeah, i meant the csv -> NS adapter
can we reuse some of your openaps -> NS stuff?
i assumed carelink csv is quite different from carelink stick comms
Ben West
@bewest
Nov 06 2015 00:48
yeah, should move that out to dcalc as well
oh, I started already
Mark Wilson
@mddub
Nov 06 2015 00:49
nice
Ben West
@bewest
Nov 06 2015 00:49
might be slightly out of date
need to cut over and require it as dependency in oref0
Mark Wilson
@mddub
Nov 06 2015 00:50
carelink csv and history from stick can be parsed the same?
Ben West
@bewest
Nov 06 2015 00:50
no
well
almost
Mark Wilson
@mddub
Nov 06 2015 00:50
is this for parsing stick or csv?
Ben West
@bewest
Nov 06 2015 00:50
which is this?
the stuff in dcalc is from oref0
Mark Wilson
@mddub
Nov 06 2015 00:51
mm from dcalc you just posted
yeah, assumed it was stick
Ben West
@bewest
Nov 06 2015 00:51
so it's output from decocare
but it so happens I'm author of mmcsv also
so IIRC correctly, it's pretty close, I think
Mark Wilson
@mddub
Nov 06 2015 00:52
yeah, just need the final step of adapting to NS treatment format
Ben West
@bewest
Nov 06 2015 00:52
that's the other thing oref0 shows, every transform I'm using to prep it for NS
should go through it again and document bits of it
Mark Wilson
@mddub
Nov 06 2015 00:53
yeah. you know i have a bias for using vendor tools if they solve the problem
Ben West
@bewest
Nov 06 2015 00:53
so do I
that's why I use the stick :-)
hehehe
Mark Wilson
@mddub
Nov 06 2015 00:54
if we had mmcsv -> NS, we could say "buy a mmconnect and enable these plugins, and your pump history and sgv history will be sitting in nightscout within 24 hours, automatically"
Ben West
@bewest
Nov 06 2015 00:54
yeah, that'd be great for a lot of people
the artificial delay bothers me quite a bit
Mark Wilson
@mddub
Nov 06 2015 00:54
yeah
see "speculative" #2 :)
Scott Leibrand
@scottleibrand
Nov 06 2015 00:55
Be nice if it was configurable. They might increase the default frequency a bit, but battery impact is a concern.
Mark Wilson
@mddub
Nov 06 2015 00:55
but at that point there's no reason to just develop for RL
*not to
@scottleibrand how did medtronic react to the idea of scraping their json? was that the context of the 0.1u request?
Dana Lewis
@danamlewis
Nov 06 2015 00:56
They were positive about it
Mark Wilson
@mddub
Nov 06 2015 00:57
neat
you wouldn't think so from the way they decided to format some of it
Dana Lewis
@danamlewis
Nov 06 2015 00:59
:)
Scott Leibrand
@scottleibrand
Nov 06 2015 01:00
Heh
Toby Canning
@TC2013
Nov 06 2015 01:01
@danamlewis @scottleibrand Did Medtronic ever acknowledge your request to be given research level access to their pump?
I see, you had to sign an NDA and can't tell us 😜
Scott Leibrand
@scottleibrand
Nov 06 2015 01:03
They actually still haven't responded yet to the "bench study" request.
Toby Canning
@TC2013
Nov 06 2015 01:04
😕
Dana Lewis
@danamlewis
Nov 06 2015 01:05
They're supposedly following up. Still :)
Toby Canning
@TC2013
Nov 06 2015 01:06
I used to be optimistic about the future!
Mark Wilson
@mddub
Nov 06 2015 01:07
just wait 5 years, we'll have a cure and they won't need to follow up
Toby Canning
@TC2013
Nov 06 2015 01:07
LOL!
Ben West
@bewest
Nov 06 2015 01:08
in 2004 when I was diagnosed, I was told by a medtronic rep that within 5 years, they'd have implantable unit for me
Toby Canning
@TC2013
Nov 06 2015 01:08
A IUD?
Ben West
@bewest
Nov 06 2015 01:09
lol
implanted pump
Toby Canning
@TC2013
Nov 06 2015 01:09
Of course ;)
Ben West
@bewest
Nov 06 2015 01:09
said she'd spent time with people testing it in CA
Toby Canning
@TC2013
Nov 06 2015 01:09
Wow
I wouldn't have guessed that
Ben West
@bewest
Nov 06 2015 01:10
5 years later I tried the cgm, finally
Toby Canning
@TC2013
Nov 06 2015 01:10
Like DRI stuff?
Ben West
@bewest
Nov 06 2015 01:10
yeah, medtronic makes a lot of implanted technology
so thinking was they'd just do the same with the pump
Toby Canning
@TC2013
Nov 06 2015 01:11
That's fantastic. I'm really glad their research is so broad
Ben West
@bewest
Nov 06 2015 01:11
well, that was 2004...
was supposed to be out in 5 years
Toby Canning
@TC2013
Nov 06 2015 01:12
They bought the "best" APS algorithim and don't have it scheduled in the next two cycles of pumps
Ben West
@bewest
Nov 06 2015 01:12
looks to me like it got canned :-(, I think I saw it in their little museum
Toby Canning
@TC2013
Nov 06 2015 01:13
I understand not having it FDA aprroved this cycle, but why not in the 670?
Ben West
@bewest
Nov 06 2015 01:13
btw, had interesting experience yesterday:
blob
Toby Canning
@TC2013
Nov 06 2015 01:13
Hopefully they are pressured into releasing it sooner, just like Dexcom has been
Dana Lewis
@danamlewis
Nov 06 2015 01:14
+1
Ben West
@bewest
Nov 06 2015 01:14
I corrected while openaps did as well
Toby Canning
@TC2013
Nov 06 2015 01:14
some major low
lows
You bottomed out there for a bit
Ben West
@bewest
Nov 06 2015 01:14
nah
that's false
Toby Canning
@TC2013
Nov 06 2015 01:14
Are you running 24x7?
Ben West
@bewest
Nov 06 2015 01:15
bad sensor
Toby Canning
@TC2013
Nov 06 2015 01:15
ahhhh
Ben West
@bewest
Nov 06 2015 01:15
that's a new sensor dropping in there...
Toby Canning
@TC2013
Nov 06 2015 01:15
59 days old?
Ben West
@bewest
Nov 06 2015 01:15
lol
Toby Canning
@TC2013
Nov 06 2015 01:15
LOL
Ben West
@bewest
Nov 06 2015 01:15
but the two high temp basals
I did micro correction
and had nasty low
ended up stacking quite a bit of insulin
found myself at 80 with more than unit on board
ate most of pack of gummy bears
did the trick
but was a little scary, feel like should capture it as adverse event somehow, per best practices/FDA
Toby Canning
@TC2013
Nov 06 2015 01:17
Are you running Open APS 24x7?
Ben West
@bewest
Nov 06 2015 01:17
oh, trying to
I have one rig set up, and I bring it with me when I got out
apparently yesterday it was working when I wasn't expecting it to!
I was on a sunset cruise on a boat
Toby Canning
@TC2013
Nov 06 2015 01:17
wow
Ben West
@bewest
Nov 06 2015 01:17
but apparently I had gotten it onto my phone's hotspot, and it was working
and then around 6pm was fairly low with lot of IOB
I throw it in my bag
oh I hve more pictures...
Toby Canning
@TC2013
Nov 06 2015 01:18
You don't have an "offline" internet mode?
Ben West
@bewest
Nov 06 2015 01:18
well, I'm using NS to get my glucose in the first place
might look into moving to using it as the NS uploader for glucose as well
Dana Lewis
@danamlewis
Nov 06 2015 01:19
I would plug in to ensure offline mode works if you are OTG
Ben West
@bewest
Nov 06 2015 01:19
battery life will improve , actually
right now more interested in getting my pebble to show relevant stuff
and also maybe adding chris' time in range report to NS
Toby Canning
@TC2013
Nov 06 2015 01:20
I wish I could get our son to look at his pebble more
nice!
Ben West
@bewest
Nov 06 2015 01:21
he's not very interested in his pebble?
or he's interested but not in the NS face
Toby Canning
@TC2013
Nov 06 2015 01:21
No, he ignores all alarms
right
Ben West
@bewest
Nov 06 2015 01:21
oh, well I'm sympathetic with that
Toby Canning
@TC2013
Nov 06 2015 01:21
Loves floppy bird
Ben West
@bewest
Nov 06 2015 01:21
can you try sleep mode?
the alarms drive me nuts, I have to take it off
i use the "sleep mode" version 24/7
Dana Lewis
@danamlewis
Nov 06 2015 01:22
Are you using NS watch face?
Toby Canning
@TC2013
Nov 06 2015 01:22
yes
Dana Lewis
@danamlewis
Nov 06 2015 01:22
I much prefer custom...
Ben West
@bewest
Nov 06 2015 01:22
there's a "sleep mode" variant which never vibrates
Toby Canning
@TC2013
Nov 06 2015 01:22
I like your custom face (saw the diagram of it, nice)
Our son turns 13 in a month, so might be a bit much
Ben West
@bewest
Nov 06 2015 01:23
where's your custom face dana?
I set up all the first-person users with sleep mode
Dana Lewis
@danamlewis
Nov 06 2015 01:23
Yea. Pulls data from DIYPS page, but easy to build whatever glance able data people want. Scott really likes the offline openAPS view to see if the loop is running - similar but focused on loop data
Ben West
@bewest
Nov 06 2015 01:23
several 11 year olds, etc, several adults who complained about alarms
Toby Canning
@TC2013
Nov 06 2015 01:24
Oh, tell scott Ben and I werejust hitting on you and saying how nice your face looked
Dana Lewis
@danamlewis
Nov 06 2015 01:24
LOL. He'll see this when he lands ;) and like I did, not assume that :p
Ben West
@bewest
Nov 06 2015 01:24
lol
Dana Lewis
@danamlewis
Nov 06 2015 01:25
Point being, might be worth doing custom to see if that helps?
Toby Canning
@TC2013
Nov 06 2015 01:25
That's a good thought, maybe I could have him pick the features he wants
Dana Lewis
@danamlewis
Nov 06 2015 01:26
+1
Toby Canning
@TC2013
Nov 06 2015 01:26
He just said, "Can we make it just like floppy bird?"
Dana Lewis
@danamlewis
Nov 06 2015 01:26
And possibly some ifftt style pushover alerts so only get alerts he knows he wants to take action on
Toby Canning
@TC2013
Nov 06 2015 01:27
good plan because he always responds to his friends texts
Dana Lewis
@danamlewis
Nov 06 2015 01:27
;)
Toby Canning
@TC2013
Nov 06 2015 01:29
It was nice chatting with you both! Funny, I've been so busy, but I've missed my online friends ;)
Ben West
@bewest
Nov 06 2015 01:29
:-)
would be interesting to ask him if he wants it to alert his friends
Dana Lewis
@danamlewis
Nov 06 2015 01:29
:)
Ben West
@bewest
Nov 06 2015 01:29
if he has friends that are hip to this
Toby Canning
@TC2013
Nov 06 2015 01:30
Just this year he stopped wanting his friends to know anything about his diabetes
Mark Wilson
@mddub
Nov 06 2015 01:30
if we're talking about how not-ideal the default NS watchface is... there are alternatives in the works :)
Ben West
@bewest
Nov 06 2015 01:30
oh I see
Toby Canning
@TC2013
Nov 06 2015 01:30
ooooohhhhhh
Ben West
@bewest
Nov 06 2015 01:30
oooOOOoooo
aaaaAaaaahhh
altogether now
Dana Lewis
@danamlewis
Nov 06 2015 01:31
Nice!
Toby Canning
@TC2013
Nov 06 2015 01:31
That's very impressive. I thought "there's not enough memory"
???
You've clearly proved those liars wrong
Ben West
@bewest
Nov 06 2015 01:31
I suspect it's mostly tedious
Dana Lewis
@danamlewis
Nov 06 2015 01:31
That might be my second favorite :) (I still am a fan of having a discrete option, but this will be nice and an in between)
Mark Wilson
@mddub
Nov 06 2015 01:32
nah, it was neither tedious nor particularly data-heavy
Ben West
@bewest
Nov 06 2015 01:32
heh, ok, well apparently we've found mark's special skills then
Dana Lewis
@danamlewis
Nov 06 2015 01:32
:)
Mark Wilson
@mddub
Nov 06 2015 01:32
i guess i was lucky to have written a different pebble app a few weeks ago
so i didn't end up writing a 5k line monstrosity
Ben West
@bewest
Nov 06 2015 01:32
apparently you have a knack for this then
Toby Canning
@TC2013
Nov 06 2015 01:32
:)
Ben West
@bewest
Nov 06 2015 01:32
now we're going to want more
Mark Wilson
@mddub
Nov 06 2015 01:33
lol
well, this one, i'm hoping to have on github within the next week
Toby Canning
@TC2013
Nov 06 2015 01:33
What is the 0/1?
Mark Wilson
@mddub
Nov 06 2015 01:33
the positioning and size of things are all customizable and it adapts, so it could be made more discreet
Toby Canning
@TC2013
Nov 06 2015 01:33
Yay!
Dana Lewis
@danamlewis
Nov 06 2015 01:34
:)
:$
Mark Wilson
@mddub
Nov 06 2015 01:34
basically, when i release this, i also plan to write an essay about why the default NS watchface is borderline morally wrong
Dana Lewis
@danamlewis
Nov 06 2015 01:34
?
Ben West
@bewest
Nov 06 2015 01:34
oy, that escalated quickly ;)
Mark Wilson
@mddub
Nov 06 2015 01:34
it conditioned me to glance at it every 5 minutes, gave me no context
i felt like a pancreas
Toby Canning
@TC2013
Nov 06 2015 01:34
LOL! I wrote some of the very first code for that
Ben West
@bewest
Nov 06 2015 01:34
ah, that's the sugar surfing influence/parents I think
Toby Canning
@TC2013
Nov 06 2015 01:35
We first had it buzz with every new reading
Ben West
@bewest
Nov 06 2015 01:35
hehehe, self-use is different enough from follower-user that we need different tools
Dana Lewis
@danamlewis
Nov 06 2015 01:35
Yea, if I just wanted BG, I'd look at my CGM or the Dexcom app
Mark Wilson
@mddub
Nov 06 2015 01:35
@TC2013 to answer your question, 0/1 is how recently it has heard from the pebble/dexcom
Dana Lewis
@danamlewis
Nov 06 2015 01:35
Context. All about context. (And in my case, predictions.)
Mark Wilson
@mddub
Nov 06 2015 01:35
definitely need to change that
Ben West
@bewest
Nov 06 2015 01:36
would be nice to be able see graph/context more easily
Mark Wilson
@mddub
Nov 06 2015 01:36
but i think it's a step in the right direction, you only need a little bit of information to troubleshoot why your readings are stale, better than CHECK RIG
Toby Canning
@TC2013
Nov 06 2015 01:36
Right... It can really only be used to supplement NS
Ben West
@bewest
Nov 06 2015 01:37
yeah, for me I just need to see "minutes ago" vs now
I don't really want/need to know much else
Mark Wilson
@mddub
Nov 06 2015 01:37
in fact, i would argue if you are relying on the graph you don't need that as a value
the graph can shift to the left to indicate the lag
Ben West
@bewest
Nov 06 2015 01:38
ah, I find it hard to see the lag
Toby Canning
@TC2013
Nov 06 2015 01:38
me too
Too random
Dana Lewis
@danamlewis
Nov 06 2015 01:38
Ditto. And the Dexcom app sometimes shifts left when there's no missing data but it's being weird
Ben West
@bewest
Nov 06 2015 01:38
even on dexcom, it's hard
Toby Canning
@TC2013
Nov 06 2015 01:38
varying with age of sensor ... and sensor to sensor
Mark Wilson
@mddub
Nov 06 2015 01:39
yeah, since the dexcom uses points instead of a line it takes work to see the lag
Ben West
@bewest
Nov 06 2015 01:39
ah, good point
Toby Canning
@TC2013
Nov 06 2015 01:40
Oh Ben, I haven't thanked you in a long time, but decocare/OpenAPS has made it so that I only have to get up 1 a week or less in the night now
Jason Calabrese
@jasoncalabrese
Nov 06 2015 01:40
wow @mddub that looks great
Dana Lewis
@danamlewis
Nov 06 2015 01:40
Yay @TC2013 !
Ben West
@bewest
Nov 06 2015 01:40
wow, that's amazing
Mark Wilson
@mddub
Nov 06 2015 01:40
@jasoncalabrese <3
Ben West
@bewest
Nov 06 2015 01:40
kevin lee and I were talking about idea of "linux load average" for glucose
which would kind of provide the "need to know" and a bit of context
maybe something with largest delta in last 30 minutes?
Jason Calabrese
@jasoncalabrese
Nov 06 2015 01:41
heard of someone else making a pebble app to enter treatments
Ben West
@bewest
Nov 06 2015 01:41
5/30/60?
if there's a -1 drop for 30 + minutes...
dexcom will show flat arrow
Ben West
@bewest
Nov 06 2015 01:41
err, sorry, more than 30 minutes
Jason Calabrese
@jasoncalabrese
Nov 06 2015 01:41
haven't looked at it yet
Ben West
@bewest
Nov 06 2015 01:42
oh that's interesting
Toby Canning
@TC2013
Nov 06 2015 01:42
Any pictures?
Ben West
@bewest
Nov 06 2015 01:42
@TC2013 due to combination of roundtrip/ihawk/openaps?
Toby Canning
@TC2013
Nov 06 2015 01:43
Well roundtrip is a port of decocare mostly and Dana and Scott's pseudocode from Feb 2015
So, yes, that is the reason
I was gone on a trip last week and our son left the moto G on without plugging it in. When it runs to 0% battery it is really hard to get to turn on.
She was miserable without for the week
Ben West
@bewest
Nov 06 2015 01:47
amazing
Toby Canning
@TC2013
Nov 06 2015 01:49
Basal rates have to be somewhat on target, but there is a lot of room for forgiveness when running the APS
Ben West
@bewest
Nov 06 2015 02:07
yeah, it allows humans to be more human-like
last night, my openaps intervened nearly 20 times
placement
actually last night, the blinking light on the wifi, and the battery kept me up for hours
Mark Wilson
@mddub
Nov 06 2015 02:16
@bewest what about the beeping to indicate a temp basal?
Ben West
@bewest
Nov 06 2015 02:30
was on a boat in the bay and didn't hear it
openaps was in my bag next to me, and I got up and walked around
if it had been on my watch, I probably would have made different decisons
just make sure I'm hydrated, etc
oh, at night?
I sleep through all alarms
NS has to alarm for half hour or more for me to ack/dismiss
I wake up to lots of sounds, but diabetes devices have trained me to ignore alarms
Matthias Granberry
@mgranberry
Nov 06 2015 02:40
@mddub I agree about the morality of the NS face. :) It's been on my to-do list to make a half-way decent one for ages. I've been spending too much time w/ my kids to finish it up and plug live data into it. If all Pebbles were running 3.x, I'd just pipe a rendered list of draw commands, but the code I have packs bg, time pairs into a 16-bit value and goes to town
diabeticgonewild
@diabeticgonewild
Nov 06 2015 02:45
Testing out rechargable AAAs on Medtronic Paradigm pumps: http://i.imgur.com/mYL8RJg.jpg
Will report on experience in a couple of weeks. I do not recommend doing this without research. FWIW, I am an EE student who understands the finicky nature of rechargable batteries pretty well.
It cost me about $45 total for batteries and equipment via Amazon but it may or may not work out depending on how the tests go...
May be possible to save money on batteries...
I would wait for my reports back (which I will report back on this HERE) before making any moves or purchases
Ben West
@bewest
Nov 06 2015 02:50
has anyone experienced any git corruption since last patches?
I believe it's solved
rpi has been surprisingly reisilient
yesterday was a nightmare use case, apparently it was working entire time
John Males
@johnmales
Nov 06 2015 02:55
So with openaps do you need to sync with a remote git repo such as on github? Do you do that regularly as a cron job?
Ben West
@bewest
Nov 06 2015 03:00
you don't need to
it's one of the side-effect of choosing to design as an extension to git, that you are able to sync to a github repo
I recommend not using the master branch
git is intended to be a "stupid" content tracking tool
git apparently slang in other countries for 'a stupid person'
openaps "hacks" this intended use to track "active RAM of an artificial pancreas"
so anything your AP might do, we want to be traceable, proveable, reproducible
so I designed it around the best tracking system I could find
Dana Lewis
@danamlewis
Nov 06 2015 03:03
@bewest That's a good explanation for it.
Ben West
@bewest
Nov 06 2015 03:04
these algorithms are so politically controversial
I wanted to remove "stupid" misconceptions easily... "what version is running, how do I update, how do I track this?" etc...
wanted to give us the space to really focus on design and making, not on janitorial work
thanks @danamlewis, I need to do a better job communicating some of the design inputs here
John Males
@johnmales
Nov 06 2015 03:10
@bewest thanks - so syncing to eg github is an optional step?
Ben West
@bewest
Nov 06 2015 03:10
it's an extra step that is not needed
nothing in it assumes github
in fact, bitbucket offers free private repos
so apparently some people are using that
but with the oref0 stuff, I'm uploading to NS
so it can be used purely as NS uploader
maybe we need an openaps-ns-monitor project, just to do the monitor part
as an nightscout-uploader
Chris Hannemann
@channemann
Nov 06 2015 03:12
+1 to that, Ben; I think that'd be a great intro for many people
@diabeticgonewild for what it's worth, I've been keeping stats on my pump's battery on the loop: https://gist.github.com/channemann/0a81661b78703fcb8da6
Ben West
@bewest
Nov 06 2015 03:13
along those lines, should we build our own gitbook, and not rely on that service?
Chris Hannemann
@channemann
Nov 06 2015 03:13
I've got a rechargeable in there right now, courtesy of @mddub; should die this evening
Ben West
@bewest
Nov 06 2015 03:14
we can build our own html using templates we control, that's generally the most robust
John Males
@johnmales
Nov 06 2015 03:15
OK that makes sense. I have do have a private repo on bitbucket because I didn't want to inadvertently advertise the pump serial number. One question - I tried to clone the repo to another rpi as a way to set it up and openaps complained when I tried to run commands on the new rpi eg openaps use pump model - am I missing a step somewhere?
Ben West
@bewest
Nov 06 2015 03:15
I've been playing with rayovac (non recharge) and they are awful
there was a change recently in openaps
I probably should have bumped a minor number
if you installed a recent version of openaps
when you add device, it's in a new ini
not tracked by git
if you clone your "instance"
I expect the openaps commands to work
if all the config is present
was there an error?
you can add me, bewest on bitbucket if you sync there and want me to look
John Males
@johnmales
Nov 06 2015 03:30
sure I'm not at home at the moment but will do. so the second rpi will run "openaps" with no error, but any openaps command in the directory with the cloned ini file etc throws errors. Sorry not at home will pass on more info later.
John Males
@johnmales
Nov 06 2015 03:48
@bewest have added you to the repo
Ben West
@bewest
Nov 06 2015 06:24
been trying rayovac and they are not so good
diabeticgonewild
@diabeticgonewild
Nov 06 2015 09:23
Cool @channemann !
I have some big news to share!!!!!!
My differential equation solver, when I plug in test values, WORKS (based off of all of the verification I have done so far!). I get the all of the "right" values, within a reasonable range!
I haven't updated my differential equation solver here yet, but I know I can definitely program a real nonlinear model predictive controller for an artificial pancreas on my own now: https://github.com/diabeticgonewild/Artificial-Pancreas-2015-Mathematical-Model/blob/master/ODESystemSolve.m
diabeticgonewild
@diabeticgonewild
Nov 06 2015 09:29
"Proof" I did it! (Although you won't be able to make sense out of this!!!): http://i.imgur.com/cJ2vV2l.jpg
diabeticgonewild
@diabeticgonewild
Nov 06 2015 09:55
Well there is stuff I am going to have to go back and "trace"...it looks like things are only going to get better with respect to this project!!!
The last two equations I am weary about but otherwise it looks great!!!!!
So 2/10 that I am concerned about...not bad.
diabeticgonewild
@diabeticgonewild
Nov 06 2015 10:33
FWIW, I know where the problem is coming from in 2/10...
Scott Leibrand
@scottleibrand
Nov 06 2015 13:50
Nice, congrats!
diabeticgonewild
@diabeticgonewild
Nov 06 2015 13:57
Thank you so much!
diabeticgonewild
@diabeticgonewild
Nov 06 2015 14:22
And backing up my stuff to GitHub right now before my computer upgrades to Windows 10!
Peter Miller
@ochenmiller
Nov 06 2015 14:23
Has anyone ported the full iob logic (including accounting for bolus and temp basal) over to nightscout? I like seeing all the info from the pump but would like iob to be reflective of openaps.
Also, be careful how you upload records from the pump! I accidentally loaded the low reservoir warnings, and pump prime to NS as bolus info! Freaked out when I saw "20U" on the timeline. Took me a moment to realize what had happened.
Peter Miller
@ochenmiller
Nov 06 2015 14:29
@bewest - regarding wifi light etc, I altered a second RPi enclosure to cover the carelink, wixel, and wifi stick. Barely any light escapes! The whole thing velcros to either the battery, or the side of the nightstand.
Scott Leibrand
@scottleibrand
Nov 06 2015 14:29
Heh. Net IOB calculation based on temp basals is next on the list for NS, but hasn't been coded yet. It will probably be based on the way we did it in my old wip/iob-cob branch though, and share iob.js code with oref0 via the NS "dcalc" stuff.
Peter Miller
@ochenmiller
Nov 06 2015 14:31
thanks @scottleibrand ! If I hack it up from your openaps code, I'll post. Right now I'm doing a lot of mental math looking at the ns timeline. Keeps my mind active! Or stressed.
Milos Kozak
@MilosKozak
Nov 06 2015 14:32
looks like NS got a new dimension with all new stuff. many people are installing just to collect data from other sources
Scott Leibrand
@scottleibrand
Nov 06 2015 14:35
:-)
Tim Howard
@tghoward
Nov 06 2015 14:39
@ochenmiller what is the wixel for? Adding additional wireless capabilities?
Milos Kozak
@MilosKozak
Nov 06 2015 14:39
wondering if this will heading to processing all data inside NS and offer servicies for other apps
Peter Miller
@ochenmiller
Nov 06 2015 15:56
@tghoward - we use the wixel to read records from the dexcom g4 transmitter
Tim Howard
@tghoward
Nov 06 2015 16:10
Ah, ok, thanks. Cool!
diabeticgonewild
@diabeticgonewild
Nov 06 2015 17:37
@ochenmiller I have to recreate records
for
apidra, Novolog, and Humalog for various glucose clamp studies, along with various pump fittings using cubic hermetian polynomial fits per @mgranberry recommendations that I did early on in the Google OpenAPS-dev group, so that everything fits to @loudnate 's tools
Peter Miller
@ochenmiller
Nov 06 2015 18:44
@MilosKozak - I have been sticking to the idea that all items necessary for the closed loop are derived and calculated locally. That way the system is less likely to have bad info injected from another source, or cease to function if off net. That said, I like the idea of parallel processing of data in NS!
Scott Leibrand
@scottleibrand
Nov 06 2015 18:47
+1. Nothing critical path for the loop should require NS, although ours can pull BG data from NS if CGM isn't plugged in.
And then everything reports to NS when online.
Jason Calabrese
@jasoncalabrese
Nov 06 2015 18:49
think I have the sets and tubbing figured out to swap between animas and mm, the clip on the silhouette is the same as the inset 30
just throw away the silhouette set and use the tubing and the normal mm cartridge
funny that I've had all the parts sitting here and didn't think they'd work
time to get everything updated, I'm the only thing in the way of making it work now
Milos Kozak
@MilosKozak
Nov 06 2015 19:03
@ochenmiller i was not thinking about ns like we have now when we started to use it. now is inconceivable reality. and when a pump with remote control becomes reality too we can do miracles
Peter Miller
@ochenmiller
Nov 06 2015 19:18
@MilosKozak , makes sense. So, more like a platform. On openaps it would be baked right in.
@jasoncalabrese , woot woot!
Jason Calabrese
@jasoncalabrese
Nov 06 2015 19:19
need to get some more supplies, but I have have 1 set now that will work
Peter Miller
@ochenmiller
Nov 06 2015 19:24
Regarding pulling CGM data from NS, I was doing that for a bit, but got into trouble when I had a mismatch with clocks. As we aren't using the dexcom receiver at all (just wixel), we now set the clock exclusively from the pump, and reference everything to that.
Biggest pain point for us is having to hand enter calibration MBG into the Dexcom receiver. I'm looking forward to G5 with the hopes that we won't have to hand carry any MBG data at all.
Well, perhaps not the biggest pain point!
Dana Lewis
@danamlewis
Nov 06 2015 19:27
awesome @jasoncalabrese !
Peter Miller
@ochenmiller
Nov 06 2015 19:56
Feeling bad. I stole @danamlewis's traditional "woot woot".
Jason Calabrese
@jasoncalabrese
Nov 06 2015 19:56
ha
Dana Lewis
@danamlewis
Nov 06 2015 19:56
;) it's not proprietary
Peter Miller
@ochenmiller
Nov 06 2015 19:57
:-)
Toby Canning
@TC2013
Nov 06 2015 20:43
😀 on the battery topic. Energizer Lithium Ultimates run 21 days if you use the APS for overnights only with the 722
Peter Miller
@ochenmiller
Nov 06 2015 20:59
@TC2013 - I've also been impressed how much life the 722 pump gets out of one AAA Lithium battery. We're running 24/7 and also getting really good battery life. I don't get the feeling that the openaps chatter adds that much drain.
Dana Lewis
@danamlewis
Nov 06 2015 21:00
Hmm. surprises me, wondering if my 522 is just…special. on all fronts.
tempted to swap to another 522 to test & see if it improves range w/ carelink and battery life
Peter Miller
@ochenmiller
Nov 06 2015 21:02
@danamlewis - our range w carelink is terrible! Though, no worse than anyone else's here as far as I can tell. Occassionally we get the "miracle connection" through two floors of the house.
Dana Lewis
@danamlewis
Nov 06 2015 21:03
Yea. But i'd even take battery improvement, get tired of changing every 3 days or so!
Peter Miller
@ochenmiller
Nov 06 2015 21:03
3 days?! Yeah, somethings's up.
Scott Leibrand
@scottleibrand
Nov 06 2015 21:03
we should get some lithiums to test then. just have to beware of the rapid drain at the end, and change it as soon as it reads low
3 days is with traditional alkalines
Dana Lewis
@danamlewis
Nov 06 2015 21:04
that sounds like a good idea, battery fairy ;)
Scott Leibrand
@scottleibrand
Nov 06 2015 21:04
:-p
Chris Hannemann
@channemann
Nov 06 2015 21:06
I got 10 days or so with lithium 24/7 looping on a 723, so that seems to be about in line with your numbers, @TC2013
Peter Miller
@ochenmiller
Nov 06 2015 21:06
We've gone a couple of days with it complaining about low battery. Not that I'd advocate that...
Chris Hannemann
@channemann
Nov 06 2015 21:06
I get about 4 - 6 on alkalines
Dana Lewis
@danamlewis
Nov 06 2015 23:33
Spurred by the convo with @TC2013 @mddub yesterday (yay for work on new watchfaces), wrote up more about my take on customizing watch faces, and another "yay!" for @MilosKozak @jasoncalabrese et al for basal visualizations that are awesome sauce: http://diyps.org/2015/11/06/the-power-of-visualizing-your-data-your-way/
@mddub wasn't sure if you had shared your watchface image/work publicly yet, so didn't link it, but if it's public and you'd like it, let me know and I can go back and add it
Jason Calabrese
@jasoncalabrese
Nov 06 2015 23:49
nice