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

21st
Jan 2018
Chris L
@woodduckie
Jan 21 2018 00:06
Thanks @scottleibrand. I've read through the determine-basal code for UAM (though didn't follow much of it...). My basic understanding was that UAM would look for unexplained rises in BG & assume it was a result of unrecorded carbs. We record every last carb & often see UAMpredBG drop off rapidly, where the COBpredBG shows a more realistic projection. Feels counter intuitive vs. what I've read in the docs/blog.
philipgo
@philipgo
Jan 21 2018 00:12
@tghoward It is in the docs because of all the problems reported here :smile:
Tim Howard
@tghoward
Jan 21 2018 00:14
@drnoname82 Okay, thanks! I'll avoid it. Was just coming around to not feeling brave ... Appreciate the reply.
Scott Leibrand
@scottleibrand
Jan 21 2018 00:33
@woodduckie when you enter all carbs, the UAM prediction is useful for situations where the carbs “disappear” into the liver or similar. The UAM prediction is highly sensitive to drops (or flattening) in the CGM readings, which is good for preventing SMB from continuing to dose too long.
bsaunder2002
@bsaunder2002
Jan 21 2018 01:02
Anyone know maximum working rom on 722?
Eric
@ecc1
Jan 21 2018 01:04
firmware version? all x22s are loopable AFAIK
bsaunder2002
@bsaunder2002
Jan 21 2018 01:07
I hope so, I thought it was <2.4 and just found out the one I craigslisted is 2.4A
Jon Cluck
@cluckj
Jan 21 2018 01:09
:thumbsup:
bsaunder2002
@bsaunder2002
Jan 21 2018 01:11
cluckj? is your thumbs up confirming 2.4A 722 will work with OpenAPS? (Im going to look in google now but can never find the pump that lists the models and supported roms)
live4sw
@live4sw
Jan 21 2018 01:12
i believe 2.4A x22 works
bsaunder2002
@bsaunder2002
Jan 21 2018 01:12
nevermind (but thanks) found the site: https://openaps.readthedocs.io/en/latest/docs/Gear%20Up/pump.html 2.4A is good to go with OpenAPS. Got the parts...where is my soldering iron?
sweet thanks!
Jon Cluck
@cluckj
Jan 21 2018 01:14
it will work
Dana Lewis
@danamlewis
Jan 21 2018 01:15
Soldering not required ;) depending on the hardware choice. But 👍🏻 for finding a pump and being ready to go!
Jon Cluck
@cluckj
Jan 21 2018 01:15
that reminded me to check my local CL for pumps, and I discovered that they made purple ones. Which I now need
bsaunder2002
@bsaunder2002
Jan 21 2018 01:19
thx...I'm going to go work on putting the pieces together. (I'll review need for soldering) Got 722+dexcom+xdrip parts+slice of radio or something+battery+charger circuit+orangepi zero/debian...going to piece that together now. Will be doc'ing / photoing in case it helps anyone. Going to drop off now. thx
Dana Lewis
@danamlewis
Jan 21 2018 01:20
@cluckj those are THE best :smile:
Jon Cluck
@cluckj
Jan 21 2018 01:21
ikr, the one I saw was a 723 3.0
jzycom123
@jzycom123
Jan 21 2018 01:36
pebble/edison/pancreable working on a few rigs。help?
Eric
@ecc1
Jan 21 2018 02:05
@bsaunder2002 are you really using an Orange Pi or was that a typo? Raspberry Pi Zero W is only $10, and I think it will be worth that much to stay with the main line of openaps development.
Scott Leibrand
@scottleibrand
Jan 21 2018 02:14
@jzycom123 pancreabble only supports pairing to one rig per watch.
That’s actually a limitation of the watch: it can only be paired to one thing at a time (normally a phone).
Peter Miller
@ochenmiller
Jan 21 2018 02:45
@jzycom123 , happy to help. I have pebble/edison/pancreable working well and reliably. Last post I saw from you, it look like you'd gotten through to the point of pairing the edison and pebble watch, still stuck?
@ecc1 , happy to report I have cc2500 transmitting fully formed G4 packets which commands such as: ./txtest -txid 63KDG -bat 216 -seq 27 -raw 173984 -filtered 161888 producing output from wixel like: 63KDG 173984 161888 216 -46 27 packet like: FF FF FF FF B0 CD 61 00 3F 03 6C BC AD C3 C9 D8 00 1D
Peter Miller
@ochenmiller
Jan 21 2018 02:51
@ecc1 next step before PR is organize a bit more. Packets will need to include seq number. Also, need to error check, on valid transmitter ID, and valid raw and filtered (not every number can be marshalled as 13-bit mantissa, 3-bit exponent) :wink:
Also, have only tested packet being received by wixel, which I think is way more forgiving than dexcom receiver...
Tim Street
@tim2000s
Jan 21 2018 03:02
@bsaunder2002 my 722 is a 2.4A firmware and happily loops.
Eric
@ecc1
Jan 21 2018 03:29
@ochenmiller nice work, looking forward to adding it
Jane49
@Jane49
Jan 21 2018 08:16
Hi
Am trying to deal with excessive zero temping after meals resulting in high. Was going to try minGuardBG-without-UAM ....
cd ~/src/oref0 && git fetch && git checkout minGuardBG-without-UAM && git pull && npm run global-install
and am getting the return - error: pathspec 'minGuardBG-without-UAM' did not match any file(s) known to git
Any suggestions? Thanks
Jane49
@Jane49
Jan 21 2018 10:34
Just realised that the above would have been included in the 0.6.0 release in November .... oops .... need to keep looking.
Scott Leibrand
@scottleibrand
Jan 21 2018 16:49
Often that is caused by too-high CR or too-low basal at mealtime. What does autotune say?
Tim Street
@tim2000s
Jan 21 2018 17:35
Just reinstalling Dev, as have finally got some time.
If I want to set-up the local web-page, is there a preference that needs to be added, or is it all standard?
Scott Leibrand
@scottleibrand
Jan 21 2018 18:28
I think it’s default. Check your crontab.
mkericson
@mkericson
Jan 21 2018 18:55
After having my rig down for 3 days - I'm finally back in business... :smile:
I tried the automated NS update tool some time ago and thought it worked. But my deployed version still indicates 0.9.2 - When I did the automated update I'm not so sure it "deployed". When I run the automated NS update now it says it up to date. As I'm learning more and more - I'm guessing that the branch or fork ?? is updated but not deployed ? Could that be the case?
This message was deleted
mkericson
@mkericson
Jan 21 2018 19:00
Here is a screen shot and what I have noticed - even before - the rig went down is that is shows "Not enacted" but its issuing temp basal and is looping. thinking this is a NS issue. Any thoughts
Scott Leibrand
@scottleibrand
Jan 21 2018 19:05
Generally that means you are running an old version of NS that is expecting received to be mis-spelled as recieved.
mkericson
@mkericson
Jan 21 2018 19:07
image.png
first post had the address on there :worried:
Jane49
@Jane49
Jan 21 2018 19:08
@scottleibrand Hi, thanks for replying. Autotune keeps raising the ISF value far above the amount I need, carb ratio can vary by 0.2 either way to the amount used before OpenAPS. My results are not as good as previously, initially with 0.5.2 they seemed better, then there were various rig and pump problems, then 0.6.0 + ill health, then F40H and the nback on humalog. Basal is good again but eating is more difficult unless it is low carb. Pre OpenAPS I put in about 10g worth of insulin roughly an hour before eating and then applied 'eating soon' with an omnipod + cgm - the results were good. On OpenAPS - a low temp target an hour before before eating and then a similar method putting in the full no of carbs with bout 30g worh of insulin when I eat and leavve the rig to deal with the remaining carbs the rig zero temps very early on for far too long and consequently by BG rises too high. I am dithering now, have changed settings, I did go with autotune settings but it became more difficult and so reerted back t ot the values I thought used to work. I can't seem to get around it and am even considering popping 10g in with a pen before a meal and retrospectively entering that value via nightscout a while into eating. I'm afraid at the moment my methods etc are rather haphazard. I can't seem to get anywhere near the flat lines others have.
Things have improved since reducing the numbers for ISF. Basal is almost identical to autotune and works well,
mkericson
@mkericson
Jan 21 2018 19:09
So is the fix to re-deploy the NS branch
Ebgineer
@Ebgineer
Jan 21 2018 19:15
@Jane49 I've had some similar issues with zero temps sooner and longer than I thought were best. My impression is that OpenAPS doesn't know about the carbs to be eaten in the future, so it calculates that you're going to be low and tries to avoid that by issuing the 0 temp basals. I've considered entering the carbs before I eat them, at the same time as giving myself the early bolus to prevent this, but haven't tried it yet.
Jane49
@Jane49
Jan 21 2018 19:25
@Ebgineer My expectation of 'eating soon' was that it allowed my BG to go down, certainly towards, the target set. I find it can zero temp before the eating soon time ends meaning that the amount of insulin peaking when I start eating isn't enough and I start rising pretty quickly.
I can get good results with low carbing, smb with cob; not too sure about smb with temp .... haven't been sufficiently regimented in trying. (I am lazy!)
If you enter carbs and insulin with only insulin given will OpenAPS go with the lack of BG rise and give a temp basal to compensate?
Ebgineer
@Ebgineer
Jan 21 2018 19:27
I don't use eating soon, but I thought it just lowered your target. In that case it would still issue 0 temp basal rates if the predicted BG was low, whether it had actually fallen below the (temporarily lower) target yet or not.
Jane49
@Jane49
Jan 21 2018 19:30
@Ebgineer Can't explain really, I do know that even when my bG is clearly going to sit at 10 for some time it can still be zero temping. I would expect the reverse, a raised basal if it is rising to fast and then zero later to claw back the balance of expected.
Tim Street
@tim2000s
Jan 21 2018 19:45
@Jane49 you have to remember that it’s not looking at the fact you are rising now, but at the IOB and where that will result in your minBG being, then deciding what insulin to give as a result of that.
Whilst there’s also the deviation versus what it expects based on the inputs received.