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

2nd
Nov 2015
diabeticgonewild
@diabeticgonewild
Nov 02 2015 02:18
Finished programming my system of differential equations solver. I wouldn't put it "up to the test" though. I just don't feel well tonight or I would be testing it like crazy....
https://github.com/diabeticgonewild/Artificial-Pancreas-2015-Mathematical-Model/blob/master/ODESystemSolve.m
Also need to comment it some more...
diabeticgonewild
@diabeticgonewild
Nov 02 2015 02:29
Pray to the math and diabetes gods for me guys...that this doesn't take too much work for it to get "correct" values.
Scott Leibrand
@scottleibrand
Nov 02 2015 03:08
@bewest as I just noted in the PRs, I think our oref0 branches are ready to merge into dev
Ben West
@bewest
Nov 02 2015 07:21
nice
Scott Leibrand
@scottleibrand
Nov 02 2015 07:23
Just freshly built three oref0 implementations: takes twice as long to image the SD cards as to install and configure everything. ;-)
Ben West
@bewest
Nov 02 2015 07:24
hmm
did you happen to try the --config switches that print openaps registration commands?
I'm looking through prs and stuff
looks like lot of it is in dev, lot of is all stuff we discussed
if you feel good about upload, I do
maybe another check to bail out on null LAST_TIME?
Scott Leibrand
@scottleibrand
Nov 02 2015 07:26
The other way I had it, with model as the last optional arg, it would take model as the last_time arg if last_time was empty.
Ben West
@bewest
Nov 02 2015 07:26
ah, I like history and model provided together
Scott Leibrand
@scottleibrand
Nov 02 2015 07:27
I haven't played with --config yet, since I have my own setup script.
Ben West
@bewest
Nov 02 2015 07:27
idea is you would use --config in your script
it's a tool's tool
you can pipe the output into | bash in your script
or we can do another curl ... | bash foo
Scott Leibrand
@scottleibrand
Nov 02 2015 07:28
I already have something working, so haven't taken the time to try to optimize further yet.
Ben West
@bewest
Nov 02 2015 07:28
does your pi hop wifis well?
Scott Leibrand
@scottleibrand
Nov 02 2015 07:29
Yeah, the wifi config is manual, but it will join home or hotspot
Ben West
@bewest
Nov 02 2015 07:29
so if you'r at home
and leave with iphone "hot"
what is supposed to happen?
mine seems to work if I boot pi with iphone on the settings/hotspot screen
then I can see one connection pop on on iphone, often
Scott Leibrand
@scottleibrand
Nov 02 2015 07:30
I usually leave home then launch hotspot like that.
Ben West
@bewest
Nov 02 2015 07:30
ah, ok
Scott Leibrand
@scottleibrand
Nov 02 2015 07:30
You can set hotspot to higher priority than home
so it will switch.
Ben West
@bewest
Nov 02 2015 07:31
so, "launching hotspot" does require some interaction on iphone?
Scott Leibrand
@scottleibrand
Nov 02 2015 07:31
But I don't bother.
Yes. iPhone limitation.
Ben West
@bewest
Nov 02 2015 07:31
like, iphone's behavior changes depending on your "current screen?'
Scott Leibrand
@scottleibrand
Nov 02 2015 07:31
yep
Ben West
@bewest
Nov 02 2015 07:31
specifically if you want a device to join, it needs to be... in settings app?
Scott Leibrand
@scottleibrand
Nov 02 2015 07:31
you have to sit in hotspot screen until something connects.
Ben West
@bewest
Nov 02 2015 07:31
or specifically on hotspot screent
Scott Leibrand
@scottleibrand
Nov 02 2015 07:31
Ya
Ben West
@bewest
Nov 02 2015 07:31
ah ok
great
wasn't sure
guess I could have googled
"is it my pi or my iphone?"
lol
Scott Leibrand
@scottleibrand
Nov 02 2015 07:32
iphones can trigger hotspot on other iPhones using BLE
Ben West
@bewest
Nov 02 2015 07:32
ok, ipad too? maybe buggy
Scott Leibrand
@scottleibrand
Nov 02 2015 07:32
so might be possible to emulate that.
Ben West
@bewest
Nov 02 2015 07:32
ok that explains a lot
Scott Leibrand
@scottleibrand
Nov 02 2015 07:32
But I think it requires same Apple ID
so may be proprietary
Ben West
@bewest
Nov 02 2015 07:32
also buggy
it fails at least 40% of time for me ipad-mini - iphone
Scott Leibrand
@scottleibrand
Nov 02 2015 07:33
Yeah, none of their BLE stuff is very reliable
airdrop either
Ben West
@bewest
Nov 02 2015 07:34
what are you folks using for insulin action again?
Scott Leibrand
@scottleibrand
Nov 02 2015 07:34
3h currently
yeah, I'm at 4h
but last 12 hours or so, feels like shorter time to envelope
some of those lows are false
the flatline lows are mostly false
but, I wonder if there's a change that takes place
once you cross like 70%+ time in threshold
there's been studies that show highs impact further highs, lows impact further lows
also, I had some icecream few hours ago when I was 70
think it was 50g maybe, but it's a slow curve
earlier I had gummy bears and saw +22
what is she using for rescue glucose?
Scott Leibrand
@scottleibrand
Nov 02 2015 07:42
Her current favorite is Starburst, cause they're 4g each and individually wrapped, so easy not to overcorrect
Ben West
@bewest
Nov 02 2015 07:55
the gummy bears take too long and have too much impact
easy to overcorrect because it takes too long
glucose tabs work great with openaps, but they're specialty
Matthias Granberry
@mgranberry
Nov 02 2015 07:56
SweetTarts are almost 100% dextrose
very quick, especially if I drink some water
Ben West
@bewest
Nov 02 2015 07:57
interesting, I found sour patch variety are faster, will look at sweettarts
starburst I've used for long time, pros/cons
I find them hard to unwrap if I'm actually low
maybe pop a few at 90 and done though
Matthias Granberry
@mgranberry
Nov 02 2015 07:58
that's my complaint about them too.
Ben West
@bewest
Nov 02 2015 07:59
and if stale they take awhile to consume too
nice, looks like openaps will handle my ice cream
Dana Lewis
@danamlewis
Nov 02 2015 08:08
Skittles work well for super low lows, but soooo easy to overdo it unless you have the snack size packages
Ben West
@bewest
Nov 02 2015 08:08
yeah, I learned to look for those
halloween is great for that
Dana Lewis
@danamlewis
Nov 02 2015 08:09
Yea :)Just found a large roll of sweet tarts
af the store today, they might work better than the tiny rolls
Matthias Granberry
@mgranberry
Nov 02 2015 08:10
I find counting sweet tarts hard when really low :( But it's rare that it's a problem.
Ben West
@bewest
Nov 02 2015 08:10
maths is hard, especially when low
Dana Lewis
@danamlewis
Nov 02 2015 08:14
Ah, not sweet tarts, I meant I found big roll of smarties. Big roll, almost necco wafer size, rather than the teeny annoying plastic wrap rolls that let the candy get away easily when you're low ;p
Ben West
@bewest
Nov 02 2015 08:14
nice
or crush in bag
Matthias Granberry
@mgranberry
Nov 02 2015 08:14
ah yeah, smarties are nice.
Dana Lewis
@danamlewis
Nov 02 2015 08:15
Yea, anything I have to count I try to stay away from, otherwise I always overdo them
Matthias Granberry
@mgranberry
Nov 02 2015 08:15
my wife likes orange soda. That stuff always sends me hyper. It always amazes me that a 20 oz thing of soda can have more carbs than a plate of pasta.
Ben West
@bewest
Nov 02 2015 08:16
yeah, amazing how my tastes in food have changed
do you find it hard to eat when hyper?
Matthias Granberry
@mgranberry
Nov 02 2015 08:17
I don't really remember the time before :-/ Sugar soda has always been verboten.
eh, I'm usually hungry when hyper.
Ben West
@bewest
Nov 02 2015 08:17
ah, sometimes I feel hungry, but hard to eat or can't decide what to eat
it's a faux-hunger, weird thingy
Dana Lewis
@danamlewis
Nov 02 2015 08:20
Depends on how high is high, and for how long
Ben West
@bewest
Nov 02 2015 08:20
yeah
context matters a lot
blob
Dana Lewis
@danamlewis
Nov 02 2015 08:21
Will be curious to see if other peoples sensitivities get tweaked as mine has from openAPS. Ie I can tell the difference when I've been 150 vs 120 overnight. Used to be 180 vs 120. And now a 250 feels like a 400
Matthias Granberry
@mgranberry
Nov 02 2015 08:21
ketones make things weird. TBH I'm not very high very long unless I eat pizza or get sick. Then things get tricky.
Ben West
@bewest
Nov 02 2015 08:21
right
Matthias Granberry
@mgranberry
Nov 02 2015 08:22
@danamlewis after a week of 70-150 I feel like utter crap when I'm outside the range.
Ben West
@bewest
Nov 02 2015 08:23
yeah
but does it correspond/correlate to insulin sensitivity/DIA or time-to-peak
Matthias Granberry
@mgranberry
Nov 02 2015 08:26
when you say "sensitivities get tweaked" do you mean insulin response or just "ugh factor" from highs?
Ben West
@bewest
Nov 02 2015 08:27
I guess there was some activity with a certain nephew not recorded
Dana Lewis
@danamlewis
Nov 02 2015 08:28
Ugh factor. Feeling more sensitive to changes in BG
(Not changing ratios. I need a better word clearly )
Ben West
@bewest
Nov 02 2015 08:29
well, I teed up that confusion
sensitive is a good word
when I talk to parents, I always advocate asking how their kid feels
rather than getting upset at 250 or whatever
I started talking through how it made me feel, and several kids chimed in that it was right; this was new information for the parents
"you never said that/complained before"
Matthias Granberry
@mgranberry
Nov 02 2015 08:31
yep. Judgement doesn't help. Feeling better does.
Ben West
@bewest
Nov 02 2015 08:31
poor kid was constantly laying down
not sleepy, but acting tired
I know that
Dana Lewis
@danamlewis
Nov 02 2015 08:32
Ben, remind me how old you were when dx'd?
Ben West
@bewest
Nov 02 2015 08:32
21 I think
or 22
Matthias Granberry
@mgranberry
Nov 02 2015 08:33
I was trying to say it's good for the parents to understand that. Long-term consequences and 12-year-olds don't necessarily get along, but they understand feeling like crap. Parents can speak to immediate needs to get motivation and motivate themselves if there is excessive fear of hypos.
Dana Lewis
@danamlewis
Nov 02 2015 08:34
Yea
Ben West
@bewest
Nov 02 2015 08:35
I've met a few preteens with fear of hpos
Matthias Granberry
@mgranberry
Nov 02 2015 08:35
but my control has actually been best since abandoning thinking about numbers and instead just playing it with intuitive adjustments based off of AGPs.
Ben West
@bewest
Nov 02 2015 08:35
yeah, insert "sugar surfing" here ;-)
I was phone with jorge valdez thursday
asked him about "why one number?"
why not show cloud/range
apparently it's because medtronic did first cgm, and did it with single value
it's like "log vs linear" for cgm
Matthias Granberry
@mgranberry
Nov 02 2015 08:36
yeah. That's how I tend to do it too. It's all a guess anyway, why not admit it? Small adjustments is how you get any other dynamic system to be stable.
Dana Lewis
@danamlewis
Nov 02 2015 08:36
Hm
Ben West
@bewest
Nov 02 2015 08:37
they had all kinds of prototypes, medtronic shipped with single number and everyone followed suit
Matthias Granberry
@mgranberry
Nov 02 2015 08:37
that's interesting, and unfortunate
Ben West
@bewest
Nov 02 2015 08:37
since then, the market has confirmed that "we" value single number that always matches
Matthias Granberry
@mgranberry
Nov 02 2015 08:37
there's so little information that CGMs give when you know they have a lot more available.
Ben West
@bewest
Nov 02 2015 08:37
that's why raw is removed
btw
Dana Lewis
@danamlewis
Nov 02 2015 08:37
Ahhh
Ben West
@bewest
Nov 02 2015 08:37
with mobile/decentralized, they wanted all numbers to match
since market likes a single number...
that's what we get
so I asked him, if community was interested in giving that a shot
just like we did with log
could we do it?
sounded like that was actually a practical request for raw experimentation that tipped the scales
because single number is really a lie, just like iob-sans-temps
Dana Lewis
@danamlewis
Nov 02 2015 08:39
Yes
Ben West
@bewest
Nov 02 2015 08:40
felt like that may have helped tip scales, he said they were already thinking about that, and there would be some signal to use to show error or a range of some kind
no promises, obviously
but there's nothing in their engineering to prevent reverse engineering, wink wink
Dana Lewis
@danamlewis
Nov 02 2015 08:40
:)
Ben West
@bewest
Nov 02 2015 08:41
well, smiley, and frowney
been at this too long for that to be the answer
Dana Lewis
@danamlewis
Nov 02 2015 08:41
Yes. But I am happy with some forward progress.
Ben West
@bewest
Nov 02 2015 08:41
hopefully stayce's speech will help with this issue
Dana Lewis
@danamlewis
Nov 02 2015 08:42
Also heard they have a new vp of data access. So will be interesting to see what they do w that
Ben West
@bewest
Nov 02 2015 08:42
he also gave me some business ideas
yeah, I'm kind of curious about positions there, especially in "cloud" I guess
wow
looks like openaps just ate my ice cream
that's wild
blob
iob + meal timing might have worked out just right
was worried a bit, when I headed low after dinner
no openaps support
Matthias Granberry
@mgranberry
Nov 02 2015 08:46
nice. I must still have some residual insulin production because I don't get much of an ice cream rise.
Ben West
@bewest
Nov 02 2015 08:46
I think I must have some production as well
ice cream/yogurt I find slow steady rise
never exceeded +5
Matthias Granberry
@mgranberry
Nov 02 2015 08:47
I made sure to be hypo when insurance made me test my c-peptide because I didn't want any fights.
Ben West
@bewest
Nov 02 2015 08:47
gummy bears will do +22
lol, so lab would have to add documentation that you were low during event?
Matthias Granberry
@mgranberry
Nov 02 2015 08:47
I never do more than a single up arrow unless I drink soda for a low.
Ben West
@bewest
Nov 02 2015 08:47
yeah... the arrow drive me crazy
ugh
the arrow makes zero sense to me
i've seen it point up with negative delta
Matthias Granberry
@mgranberry
Nov 02 2015 08:48
the HL of c-peptide is very short, so if I was low there wouldn't be endogenous production and it wouldn't be in the sample.
Ben West
@bewest
Nov 02 2015 08:48
tonight, even
blob
Dana Lewis
@danamlewis
Nov 02 2015 08:49
Zero c peptide here, no production left. No luck for me if the cure first comes to those who still have production
Ben West
@bewest
Nov 02 2015 08:49
moments ago
Matthias Granberry
@mgranberry
Nov 02 2015 08:49
the arrow is an average of raw values (3 or so). They spike up fast, so sometimes it lags.
Ben West
@bewest
Nov 02 2015 08:49
when were you diagnosed, @danamlewis ?
Dana Lewis
@danamlewis
Nov 02 2015 08:50
14
Actually was never required to have c peptide for insurance purposes until this year, though, so no idea when I went to zero
Ben West
@bewest
Nov 02 2015 08:54
I've never had a test, so it's mostly uh...
well nights like tonight make me think something is going on
Dana Lewis
@danamlewis
Nov 02 2015 08:54
Wonder though if that corresponds to how people deal with carbs and GI, and/or various levels of liver dumps
Ben West
@bewest
Nov 02 2015 08:54
right, the context matters
from what I understand, ketones is few steps aways from death, and is side-effect of several chain-reactions that precede it
well, keto-acidosis
you've talked a lot about insulin pooling
for me, I fast a lot
so I'm interested in... glycogen/glucagon pool
Matthias Granberry
@mgranberry
Nov 02 2015 08:56
DKA is a nasty deal. It shows that we weren't designed.
Ben West
@bewest
Nov 02 2015 08:56
when ketones occur, I figure its like it's full/overflowing
when it's empty, it's like insulin is more effective
faster, too
I'm sure that's not right
I haven't thought/heard about pooling till I met you
but I had heard about things the liver does
which seemed to match what I felt/experience/ate
Dana Lewis
@danamlewis
Nov 02 2015 09:01
Heading to bed. Chat more later!
Matthias Granberry
@mgranberry
Nov 02 2015 09:02
Insulin pooling is one of the most frustrating parts of pump therapy.
Ben West
@bewest
Nov 02 2015 09:03
interesting
also, my family has never been part of my diabetes care
so times like this, where family is around
they've finally understood how much it impacts things I do, but they have a hard time understanding all the context
so if they see me munching some rescue carbs, they get concerned
but at that point, everything is done, it's hard to explain
"do you want to get something to eat?"
no...
just had sandwich hour ago, if I eat now I won't eat dinner...
managing appetite...
Matthias Granberry
@mgranberry
Nov 02 2015 09:08
yeah. That's frustrating with my step kids. They're so helpful, but I just want to wait it out. I was DX'd at 11, my brother at 18 mos when I was 4. I didn't think they controlled him well (lots of scary hypos and borderline negligence paired with an attempt at tight control) so he developed hypo-unawareness pretty bad and had lots of unconscious/seizure-style lows while no parents were around. It's made me fiercely independent of my own management and makes my own family dynamic a little weird.
they tried to do right by him way before the DCCT trials, but they went overboard and didn't do carb counting well because it wasn't really done back then
Ben West
@bewest
Nov 02 2015 09:10
amazing
Matthias Granberry
@mgranberry
Nov 02 2015 09:12
he was DX'd right around when the first blood glucose test strips were available
the accuracy wasn't there to do tight control for too long a time.
Ben West
@bewest
Nov 02 2015 09:16
and now within last 2 years cgm caught up to meters
when I was diagnosed, I traveled 45 minutes to dr
I remember doing the trip and thinking through "what do I want to do?"
maybe compuers would be helpful if this diabetes stuff is important
I had already been told there was a patient in CA that had implanted AP, and it would be everywhere in 5 years
medtronic rep
so I figured the computer stuff was better left to people more suited to the task, and I was more suited for arts & humanities and would continue my music studies
5 years later I was working at alexa on software professionally and realized I had to start getting involved
Matthias Granberry
@mgranberry
Nov 02 2015 09:22
I wish I'd done something earlier. I just wasn't impressed with CGMs until I tried a Dexcom G4 2.5 years ago and saw that the world had changed. Pumps are a mixed bag, but CGMs and good data analysis are world-changing.
Ben West
@bewest
Nov 02 2015 09:23
that decision took 5 years to bear fruit
1.5 years to build frustration to start working everything in public
took 2 years from then to start work for meraki, where I could learn from MIT team
the MIT team was completely unsympathetic: "just reverse engineer it and graph the data, obsiously"
took 4 years at meraki to develop decocare
took another 18 months to make it what it is now + rileylink + openaps
last 18 months is full time, watching what people do, responding, etc
I've been blessed by cisco buying meraki to be able to do that
Ben West
@bewest
Nov 02 2015 09:29
now, though... the "fruit" is new every day
Matthias Granberry
@mgranberry
Nov 02 2015 09:32
on that note it's 3:30 and I got xDrip playing (sort of) nicely with Marshmallow and I should probably get some sleep before insulin sensitivities get completely messed up tomorrow.
also fixed the redelivery attempt problems for REST uploads. I think it's been a good night.
Ben West
@bewest
Nov 02 2015 09:38
:-D
night, really glad you are enjoying REST api
I'm very open to feature requests, thinking about starting /v2/ with ranges and things
Matthias Granberry
@mgranberry
Nov 02 2015 09:41
ranges? I'd like to be able to grab all events modified since a certain time, deletions, and reworked treatments. Making a reliable and responsive (caching) rich client is difficult right now.
Ben West
@bewest
Nov 02 2015 09:42
hmm, should be possible, would love to arrange hangout and record it
if there's something missing, maybe I can figure out how to add it quickly
/v1/ is basic CRUD, read-write
I've noticed, openaps as an app requires lots of checks to be a "rich client"
an app
to my knowledge, NS is best REST api for glucose, I'd like to keep it that way
have great night :-D
Matthias Granberry
@mgranberry
Nov 02 2015 09:47
glucose is solid. Treatments are a different situation. I haven't figured out a good way to sync them that doesn't hurt bandwidth. A creationTime field would go a long way. And if modification/deletion is ever allowed, a deleted: True and a modificationTime field instead of simply modifying/deleting would make it easier to politely keep state current.
Ben West
@bewest
Nov 02 2015 09:48
yeah, those would be solid rules to add to all types
hmm, seems like 4-5 work items on NS assuming entries is same as treatments (and it's not)
it's 3-4 work items aways from being true
Matthias Granberry
@mgranberry
Nov 02 2015 09:50
treatments are also a pain to parse with GSON. They are maps vs. objects, and some of the types have changed a bit. It makes statically typed languages tricky to work with.
Ben West
@bewest
Nov 02 2015 09:50
right
that's one of the reasons I've been tiling towards simple type field with gauranteed types
I've noticed for typed, even in python, it's extra extra tricky
protobuf is nice
I'd trade protobuf all day long if people would like
John Males
@johnmales
Nov 02 2015 09:52
Hi @bewest gee your "back story" is very interesting. I didn't realise it has been such a long road. Hats off to you sir
Ben West
@bewest
Nov 02 2015 09:52
it's an even longer story for people like @mgranberry
Matthias Granberry
@mgranberry
Nov 02 2015 09:52
protobuf enforces static typing, but it's not necessarily "easy" to start working with. After getting used to it, it isn't so bad.
Ben West
@bewest
Nov 02 2015 09:52
I stand on the shoulders of giants
for people using json, it's very straightforward transform
so any logic that can work with json will work with protobuf
with the benefit that protobuf is much much easier to work with in static/compiled
Matthias Granberry
@mgranberry
Nov 02 2015 09:55
right, but the API is a little clunky. But probably no worse than JSON with the advantage that no data will come along that won't quite parse.
Ben West
@bewest
Nov 02 2015 09:55
@johnmales I love you'r story/focus
John Males
@johnmales
Nov 02 2015 09:55
@mgranberry a double hats off :)
Ben West
@bewest
Nov 02 2015 09:55
I want my home/office/car to be safe place
Matthias Granberry
@mgranberry
Nov 02 2015 09:55
I wouldn't complain about a set of .proto endpoints.
Ben West
@bewest
Nov 02 2015 09:56
blanket coverage
intersting
just json/proto support on REST?
thinking about additional "logical" endpoints on /v2/
/v2/ == /logic/
want to keep /v1/ clean for pure CRUD
no logic
Matthias Granberry
@mgranberry
Nov 02 2015 09:57
why not? You can already do /api/v1/entries.json, an entries.proto wouldn't be bad
Ben West
@bewest
Nov 02 2015 09:57
sure, there was some chatter about that
wasn't sure anyone would use it
Matthias Granberry
@mgranberry
Nov 02 2015 09:58
3 months of SGVs is ~2.5 mb data, although compressed it would be much less. I'd rather not have the overhead in a mobile app if I could avoid it.
Ben West
@bewest
Nov 02 2015 09:58
guess it's still significant even with the full http headers
John Males
@johnmales
Nov 02 2015 09:58
@bewest I agree with the safe place idea, just want to make this management as unobtrusive as possible, as I'm sure we all do
Ben West
@bewest
Nov 02 2015 09:59
if payload is compressed that way, it's still fairly large savings
@johnmales that's what I love about your story
and your focus
"my home is a safe place"
it's plain and clear, you never say it
it gives the story clarity, focus
different people need to "own" different stories
I owe you some sim cards and info on them
John Males
@johnmales
Nov 02 2015 10:00
Just not enough hours in the day, the day job is, ahem, rather busy
Matthias Granberry
@mgranberry
Nov 02 2015 10:00
for backfills it's not insignificant either. But it's probably not worth a lot if it's significant work.
Ben West
@bewest
Nov 02 2015 10:05
there's a fair bit of hackery in openaps/oref0 to figure out last treatment
so if v2 == ranges, thinking about what latest might mean for slightly different contexts
Matthias Granberry
@mgranberry
Nov 02 2015 10:06
yep. For OpenAPS an "active" state might be useful
Ben West
@bewest
Nov 02 2015 10:07
any "rich client"
Matthias Granberry
@mgranberry
Nov 02 2015 10:07
but I'd like to see a way to ensure 100% of the record has been copied
Ben West
@bewest
Nov 02 2015 10:08
my treatments post responds with all records posted succefully and which ones were inserted vs updated
Matthias Granberry
@mgranberry
Nov 02 2015 10:08
a git-style hash chain might be useful, or a Merkle tree for ensuring consistency, but that seems like overkill.
Ben West
@bewest
Nov 02 2015 10:08
git compat would be beautiful, and something I shoot for
git push clinic, or even better git push -f clinic :-D
but... automated
git just because of traceability requirements
Lori
@LoriLori
Nov 02 2015 10:18
been dx ed at 22 like Ben :)
Ben West
@bewest
Nov 02 2015 10:19
were in in college?
Lori
@LoriLori
Nov 02 2015 10:20
yes, last 2 years of college were with MDI
Ben West
@bewest
Nov 02 2015 10:20
yeah, same
ah, I spent < 9 months MDI
Lori
@LoriLori
Nov 02 2015 10:22
only last 3 weeks on non MDI, strange feeling
see this "better average/->worse feeling at high" too, several times level of feeling bad at high lowered, like acquired Dex, switch to Apidra and now pump
Ben West
@bewest
Nov 02 2015 10:24
it's different tool
means different "rules"
mean different politics
there are human values encoded in software... that's big part of why I write software instead of music/poems
I can't believe system ate that ice cream
blob
may take some gummy bears before bed
I mean this is new tool for me
this is really odd, unexpected
oh well, hope your morning goes great @LorelaiL
Ben West
@bewest
Nov 02 2015 10:30
at 2am, nearly 3, mine is clearly ruined
;-)
Lori
@LoriLori
Nov 02 2015 10:36
woke with 300s not a best morning
still work in progress with temp basal history to calc max IoB and then happy to jump on adaptive basal openAPS
most surprising is there is difference with bolus 3U imediate and bolus 3U in 30 minutes
will need to do it more often to confirm, looks very promising
Scott Leibrand
@scottleibrand
Nov 02 2015 16:02
I think the default NS+OpenAPS temp basal display should be called "city skyline" view. :-)
Jason Calabrese
@jasoncalabrese
Nov 02 2015 16:02
I like that Skyline and Icicle
the steps in the screenshot @danamlewis posted yesterday really look cool
For some reason gitter app won't let me insert that directly.
Jason Calabrese
@jasoncalabrese
Nov 02 2015 16:05
wow
Milos Kozak
@MilosKozak
Nov 02 2015 16:08
i didn't imagine basals are going to be such a toy for you :)
Scott Leibrand
@scottleibrand
Nov 02 2015 16:20
Totally! Been wanting that feature for over a year. :-)
next need to get IOB calculations to include temps, and incorporate the rest of wip/iob-cob features, including purple line
Milos Kozak
@MilosKozak
Nov 02 2015 16:22
that's i want too. is the code already in openaps?
Jason Calabrese
@jasoncalabrese
Nov 02 2015 16:23
hoping to get the calculators into https://github.com/nightscout/dcalc
then they can be shared between NS and oref0
the NS plugins would then be a wrapper around dcalc
Scott Leibrand
@scottleibrand
Nov 02 2015 16:26
Check out the old wip/iob-cob branch. It was pre-funnelcake, so not mergeable, but has all the functionality we want.
We might also want to look at loudnate's predict stuff.
does a lot of the same things, probably more cleanly. Not js unfortunately though.
Jason Calabrese
@jasoncalabrese
Nov 02 2015 16:27
parts of the AR2 plugin can be copied to add a new forecast
Scott Leibrand
@scottleibrand
Nov 02 2015 16:27
And yes, the net IOB calc stuff is in oref0 dev branch.
Dana Lewis
@danamlewis
Nov 02 2015 16:36
+1 for skyline :)
Dana Lewis
@danamlewis
Nov 02 2015 19:45
@mgranberry @bewest this is what I was talking about last night. Just used them for a low, going to be my new favorite for anything that needs >4g (a one starburst fix to ward off a borderline low for me): http://www.smarties.com/product/giant-smarties/
Matthias Granberry
@mgranberry
Nov 02 2015 20:05
I used to carry smarties around. They definitely work quickly and taste decent. The problem is I have a sweet tooth and they're too tempting. Dextrose chalk doesn't tempt me.
Dana Lewis
@danamlewis
Nov 02 2015 20:12
:) the big ones look just enough like a glucose tab to not seem like candy compared to the little ones for me. (and not nearly as annoying to get out of the roll).
Matthias Granberry
@mgranberry
Nov 02 2015 21:13
Says the person using starbursts. I'm always tempted to eat the wax paper if I'm having a bad one rather than unwrap those things.
Dana Lewis
@danamlewis
Nov 02 2015 21:13
Touche ;p
diabeticgonewild
@diabeticgonewild
Nov 02 2015 21:32
@mgranberry Starburst Minis are unwrapped and are like crack.
diabeticgonewild
@diabeticgonewild
Nov 02 2015 22:46

Well I got my ODE solver all commented and stuff. BUT I have to get the "random walks" all taken care of and figured out before I can start testing out the ODE solver with test values :( .

https://github.com/diabeticgonewild/Artificial-Pancreas-2015-Mathematical-Model/blob/master/ODESystemSolve.m

But that should only take a 1 or 2 days...if I am feeling well...
diabeticgonewild
@diabeticgonewild
Nov 02 2015 23:00
In other words I have to create a solver to solve for values used for the ODE solver
It's just for 3 values that it's needed. For the solver that I have to create.
f_sub_g_of_t, I_sub_m_of_t, and f_sub_m_of_t, are what I have to solve for with random walks...then after I solve for that, I can use that with test values in the ODE solver to test things out...
But damn low blood sugar...
diabeticgonewild
@diabeticgonewild
Nov 02 2015 23:14
Having autonomic problems makes low blood sugar in general a lot more unpleasant. First gastroparesis, which makes the lows last longer. Plus low blood pressure problems and tachycardia, which makes the palpitations from low blood sugar a lot worse.