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

6th
Sep 2017
Craig Brenner
@cbrenner
Sep 06 2017 01:17
@danamlewis Cool, thanks...didn't scroll down far enough to see this in the IFTTT integration.
PopcornPam
@PopcornPam_twitter
Sep 06 2017 01:23
I am having an issue with my ISF. In my pump it is set as 60, and I put the autotune_isf_adjustmentFraction as 0, yet the pill and logs consistantly have my isf in the 80's.
Dana Lewis
@danamlewis
Sep 06 2017 01:23
do you have autosens turned on as well?
PopcornPam
@PopcornPam_twitter
Sep 06 2017 01:25
as far as I know yes
Would that be under settings?
Dana Lewis
@danamlewis
Sep 06 2017 01:25
then that would also adjust your ISF. you should turn off autosens if you don't want it adjusting ISF, either
you can either take your autosens multipliers to 1.0 for both min and max; or delete autosens from your runagain/save/run it again
PopcornPam
@PopcornPam_twitter
Sep 06 2017 01:26
I thought putting isf_adjustmentfraction would stop that
hmm...
Dana Lewis
@danamlewis
Sep 06 2017 01:26
no, that just is for autotune to limit ISF, that's why it's called autotune_isf_adjustmentFraction
PopcornPam
@PopcornPam_twitter
Sep 06 2017 01:26
Oh.. duhhh
It does seem like less of an issue since I have started FIASP... now the issue is super long zero temps after boluses
PopcornPam
@PopcornPam_twitter
Sep 06 2017 01:35
I wonder if it is the bolus snooze that needs adjusting
Scott Leibrand
@scottleibrand
Sep 06 2017 01:49
if your eventual COBpredBG is above target, it will bolus snooze
if it's not above target, then your ratios may be off
PopcornPam
@PopcornPam_twitter
Sep 06 2017 01:49
it is showing that I am going to go super low, but I am sure that the predictor lines are incorrect
Scott Leibrand
@scottleibrand
Sep 06 2017 01:50
do the bolus calculator math yourself: given your BG, COB, IOB, and ratios, where should you end up?
if you're not sure which things to multiply/divide, feel free to post the numbers here and I can walk you through it
PopcornPam
@PopcornPam_twitter
Sep 06 2017 01:52
Right now the predictor line looks right (220, isf 60, no COB), but a re-occuring trend is that it the predictor line will jump from predicting bg of 39 to bg of 200 in 1 5 minute change
Also, the predictor line is correct for my pump ratio of 60, but the pill still is calling my isf 86, and I can't seem to get the rig to go back to using the pump set isf
Scott Leibrand
@scottleibrand
Sep 06 2017 01:56
easiest fix for that is to move your ~/myopenaps/ directory to a backup location, re-run oref0-setup from oref0-runagain, and then copy your preferences.json back over.
PopcornPam
@PopcornPam_twitter
Sep 06 2017 01:58
move it over to drill down later?
Scott Leibrand
@scottleibrand
Sep 06 2017 02:04
if you feel like it. or just chalk it up to old data files and keep the backup copy around unused in case you ever need it.
umich10
@umich10
Sep 06 2017 02:34
xDrip+/android has randomly stopped uploading to NS with the error message that it "can't do rest api". nothing has changed in the settings and we've been using xdrip forever. any ideas? BT connection from receiver to phone still live.
jquitos
@jquitos
Sep 06 2017 02:45

hello... I'm not sure what I did... but hoping didn't screw anything up. Was looking up preferences and think I went into edit-pref vs cat-pref... I didn't think I made any changes and accidentally canceled out of session. This is what popped up when I went into edit-pref again...

E325: ATTENTION
Found a swap file by the name ".preferences.json.swp"
owned by: root dated: Tue Sep 5 22:35:34 2017
file name: ~root/myopenaps/preferences.json
modified: YES
user name: root host name: johnny5
process ID: 30211
While opening file "preferences.json"
dated: Sun Jul 30 20:37:08 2017

(1) Another program may be editing the same file. If this is the case,
be careful not to end up with two different instances of the same
file when making changes. Quit, or continue with caution.
(2) An edit session for this file crashed.
If this is the case, use ":recover" or "vim -r preferences.json"
to recover the changes (see ":help recovery").
If you did this already, delete the swap file ".preferences.json.swp"
to avoid this message.

Swap file ".preferences.json.swp" already exists!
[O]pen Read-Only, (E)dit anyway, (R)ecover, (D)elete it, (Q)uit, (A)bort:

When I went into cat-pref nothing looked changed...advice? Sorry...so newb :worried:

Miiko
@tw_miiko_twitter
Sep 06 2017 02:49
SMB-Troubshooting-Treatment-in-the-future.png
@scottleibrand hello, dangerous treatment in the future is found! Where does that come from?
SMB-Troubshooting-Treatment-in-the-future.png
Miiko
@tw_miiko_twitter
Sep 06 2017 03:00
in 2021 Corretions Bolus 11.4 U?! and In 2076???
what can happen when such a correction is given in the night unconsciously...
Miiko
@tw_miiko_twitter
Sep 06 2017 03:15
anyone knows this Problem?
Miiko
@tw_miiko_twitter
Sep 06 2017 04:10
because Dev # 2782 "nightscout/0.8.2-beta2" ???
Paul Andrel
@stavlor
Sep 06 2017 04:15
0.8.2-beta2 is quite old current Nightscout master is 0.9.2 and has been that for roughly 9mo
0.8.2 is rougly October of 2015
Dana Lewis
@danamlewis
Sep 06 2017 04:22
That usually happens from a device with time set wrong; or wonky decoding. Just delete from NS and it should be fine. If it reuploads those same future entries, then it's somehow on the device, but usually you can delete from NS and that resolves everything else
Miiko
@tw_miiko_twitter
Sep 06 2017 04:26
@stavlor I have yesterday a Pull Request from Nightscout Dev, i am sure...
skubesch
@skubesch
Sep 06 2017 04:27
Hello group, I have a question about offline looping. I'm using G4-upload and haven't been able to get either rig to recognize data during the 2-hour start up (hoping it would read the raw data). Tried using an Edison/Explorer with G4 receiver plugged into OTG port while rig was being powered through UART port, and now trying with a Raspberry Pi/TI rig with G4 receiver plugged into a USB port. Logs just show that BG data is too old. Could anybody tell me what I might be missing?
IMG_2774.PNG
Screen Shot 2017-09-06 at 12.23.23 AM.png
Scott Leibrand
@scottleibrand
Sep 06 2017 04:28
@jquitos it looks like you had made some changes when you were editing the file, but then quit the editor (by disconnecting) without saving them. if you didn't want the changes you made, you can just hit D to delete the autorecovery version
@skubesch which CGM type do you have in oref0-runagain?
skubesch
@skubesch
Sep 06 2017 04:29
cat-runagain shows --cgm=G4-upload
jquitos
@jquitos
Sep 06 2017 04:30
Thank you @scottleibrand
Miiko
@tw_miiko_twitter
Sep 06 2017 04:31
@danamlewis thanks. I deleted it , but it came back still! It can not be deleted from NS!!! It came back still with a other future time, and with a other Corretion Bulos. It likes a cycle...
Scott Leibrand
@scottleibrand
Sep 06 2017 04:32
@tw_miiko_twitter sounds like your rig pulled the record from the pump that way then. you could also try deleting the pumphistory*.json files, then deleting the future-dated entry from NS, and see if that clears it up.
@skubesch that's correct...
@skubesch probably time to start looking at your glucose.json files to see if they have the glucose value from the raw data
Miiko
@tw_miiko_twitter
Sep 06 2017 04:34
@scottleibrand thank you very much! I try it, and tell you later
@danamlewis thank you very much! If it is timesetting or decoding wrong, how to fix it? Please more Information.
skubesch
@skubesch
Sep 06 2017 04:41
Screen Shot 2017-09-06 at 12.36.33 AM.png
Dana Lewis
@danamlewis
Sep 06 2017 04:41
If what scott suggested didn't work, you'll need to push the wrongly timestamped event out of history by setting a bunch of temporary basal rates until the bad event is pushed out of history.
skubesch
@skubesch
Sep 06 2017 04:41
Should I be looking at /myopenaps/cgm/glucose.json or /myopenaps/monitor/glucose.json? I think it is getting the data because even though this is the same receiver that gets raw data uploaded to NS through my android uploader phone, I now have a second set of dots in my NS site with the receiver plugged into my rig
Scott Leibrand
@scottleibrand
Sep 06 2017 04:41
@skubesch both I believe
skubesch
@skubesch
Sep 06 2017 04:47

I'm not sure how to read the timestamp but I used tails (-F) to see if new entires are being added and I'm seeing these:
monitor/glucose.json
{
"trend_arrow": "NOT_COMPUTABLE",
"display_time": "2017-09-05T23:31:04-04:00",
"system_time": "2017-09-06T06:31:05-04:00",
"direction": "NOT COMPUTABLE",
"dateString": "2017-09-05T23:31:04-04:00",
"date": 1504668664000,
"unfiltered": 108800,
"filtered": 52008,
"device": "openaps://mipi/cgm",
"rssi": 180,
"sgv": 5,
"type": "sgv",
"glucose": 5
}

and in cgm/glucose.json
{
"trend_arrow": "NOT_COMPUTABLE",
"display_time": "2017-09-05T23:31:04-04:00",
"system_time": "2017-09-06T06:31:05-04:00",
"direction": "NOT COMPUTABLE",
"dateString": "2017-09-05T23:31:04-04:00",
"date": 1504668664000,
"unfiltered": 108800,
"filtered": 52008,
"device": "openaps://mipi/cgm",
"rssi": 180,
"sgv": 5,
"type": "sgv",
"glucose": 5
}

Miiko
@tw_miiko_twitter
Sep 06 2017 04:51
@danamlewis it should be " time setting wrong" . Pls lesen loop-Log: Refreshed pumphistoryWarning: clock input Invalid Date is unzoned; please pass clock-zoned.json instead
and meal.json
how to fix it then?
Miiko
@tw_miiko_twitter
Sep 06 2017 05:15
@danamlewis @scottleibrand I have by reboot temporarily solved the problem, but not sure if it will come back any time. When it comes, I will try to solve the pumphistory.json. Thank you very much, and good Day for you.
Steve Bernacki
@sbernacki
Sep 06 2017 13:21
Hopefully simple question: to switch from autosens to autotune, I just need to re-run the setup script with autotune enabled, correct?
dan lightner
@danimaniac
Sep 06 2017 13:28
@cbrenner Just thinking out loud here, but have you looked into TeamViewer for remote access? It's cross platform and free for personal use. I use it all the time (not to access my rig) from desktop to mobile, mobile to desktop, desktop to desktop, etc. Also, it does not require a monthly hosting fee. Again, just thinking out loud.
Scott Leibrand
@scottleibrand
Sep 06 2017 14:14
@sbernacki you don't switch from one to the other. If you want to enable autotune in addition to autosens, then yes, just re-run oref0-setup.
jcwarrior
@jcwarrior
Sep 06 2017 14:37

Hi! I need some help to adjust an uncommon setup of OpenAPS for my son.

He has many emotional highs (very high, 350 md/dl and above). But we must not treat it with insulin because when emotional ends, he go quickly low (without extra insulin). He has emotional highs almost every day at school (not if we are at home) and I wish OpenAPS doesn't add insulin when he is at school (from 9:00h to 13:00h).

I have tried to reach this behavior defining a BG target range for that period from 170 to 300, but OpenAPS reports 185 mg/dl as target and then puts extra basal. Any other has a similar situation of deactivate adding insulin at certain periods of the day? I do not want to lose the ability to suspend the infusion if a low is predicted.

Thank you!!!

Dana Lewis
@danamlewis
Sep 06 2017 14:43
Do you have "adjust targets when high on"? Check preferences, turning that off may be all you need to do along with the 170-300 target for that time of day
jcwarrior
@jcwarrior
Sep 06 2017 14:46
I have override_high_target_with_low to false
Scott Leibrand
@scottleibrand
Sep 06 2017 14:46
The maximum target is 200, so setting the high end of the range above that has no additional effect. If you want a higher temp target than that, you would need to raise the low end of the target range instead.
That is a different preference. Do you have "adjust targets when high"? If so, try turning that off. It's off by default though.
^ turn that one off
jcwarrior
@jcwarrior
Sep 06 2017 14:49
thanks Scott and Dana. adv_target_adjustments: false too
maybe the best option is another rig only for the school, with a current_basal_safety_multiplier: 1
or change that multiplier automatically by crontab.... but a bit intrusive...
Dana Lewis
@danamlewis
Sep 06 2017 14:53
Yea, lowering the multipliers down might also help, or adjusting maxiob during that day, but switching that back and forth is probably not ideal, so setting a high target (i.e. 180-200, or 200-200) during school might be enough?
jcwarrior
@jcwarrior
Sep 06 2017 14:53
I have not raised the low end of target because I'd like to maintain predefined basal if he doesn't go out of range
I will test different alternatives. He goes to 350 easily, and 200-200 target will put basal 0 when he is at 180 but puts extra insulin whe go beyond 200. Then hypo comes.... :-) Better try to avoid lows at school....
Scott Leibrand
@scottleibrand
Sep 06 2017 14:55
A couple people have set up two preferences.json files and a couple cron jobs to switch between them at the appropriate times.
that would avoid having to have two different rigs with different settings and having to remember to turn off the school rig at home
jcwarrior
@jcwarrior
Sep 06 2017 14:57
oh, that sounds good. better than modify preferences. I will try with two preferences.json. THANK YOU again Dana and Scott. You are awesome again :-)
if you do not mind I'll take advantage to try to clear another doubt
My son has very differents ratios on breakfast and lunch
14chs for insulin unit at breakfast , 35-40 at lunch
The predictions painted on nightscout is by going low almost everytime whe breakfast is later than usual
The predictions painted on nightscout is by going low almost everytime when breakfast is later than usual
Could I do something to improve that prediction? I think it is for that difference on ratios, but I am not sure
jcwarrior
@jcwarrior
Sep 06 2017 15:04
(sorry for my English. it is very difficult for me to express all these ideas without making great mistakes in writing)
Scott Leibrand
@scottleibrand
Sep 06 2017 15:05
from my experience, more aggressive breakfast ratios are usually compensating for something else, such as difficulties with insulin timing. having a CR that low probably means that he will actually go low after a breakfast bolus if nothing else (like a mid-morning snack) is done to compensate. some people have had success with raising the CR and instead compensating for the big spike from eating sugary breakfast on an empty stomach by doing an eating soon (bolus or temp target) and/or an early prebolus
@bdr1177 and I just went through an extensive analysis of their situation (which sounds superficially similar) a few days ago, so he may be able to share what they've done since then and what has worked well
jcwarrior
@jcwarrior
Sep 06 2017 15:08
My son is on honey moon yet, and needs only a bit of insulin at night (basal is 0,025/h). Maybe it has any relation with the so differents ratios
Scott Leibrand
@scottleibrand
Sep 06 2017 15:09
does your ISF vary that much over the day as well? or just the CR?
jcwarrior
@jcwarrior
Sep 06 2017 15:09
just CR. ISF maintains with no big differences at 600-650
Scott Leibrand
@scottleibrand
Sep 06 2017 15:10
oh, wow. that is crazy high. he must be little. :)
jcwarrior
@jcwarrior
Sep 06 2017 15:10
3 years old
Scott Leibrand
@scottleibrand
Sep 06 2017 15:10
when you said school I was assuming at least 6
jcwarrior
@jcwarrior
Sep 06 2017 15:10
and honey moon. 1 year from debut
It's difficult to adjust all settings to make it comfortable. But I'd like OpenAPS and sure it will help us
I'm doing tests and tests and tests... to find the best settings to fight with so high sensitivity and the emotional highs
Brian Rabinovitz
@bdr1177
Sep 06 2017 15:14
@scottleibrand @jcwarrior As Scott mentioned, he helped me do some troubleshooting with a similar situation where we had a much higher breakfast CR (1:17) than other CRs (all between 1:35 and 1:45). I'm still testing it it, but had good result this morning by changing breakfast CR to 1:35 and doing ES 45 minutes prior to breakfast and pre-bolusing. This also helped prevent a pretty severe post-breakfast crash that we were experiencing about 90 minutes after breakfast.
jcwarrior
@jcwarrior
Sep 06 2017 15:18
Thank you Brian. But my son doesn't usually have lows after breakfast. I think breakfast CR is not very wrong. Maybe is simply part of the complexity of this disease...
Brian Rabinovitz
@bdr1177
Sep 06 2017 15:20
@jcwarrior That's good. We were on a bit of a roller coaster with post-breakfast high and then crashing into a low. Do you pre-bolus for breakfast? If so, how long before?
jcwarrior
@jcwarrior
Sep 06 2017 15:23
only pre-bolus if he is quit high. 10-20 minutes (25 minutes if 300 or above, not usually). If he arrives at breakfast at 150mg/dl then no pre-bolus or go low at the beginning of breakfast
Brian Rabinovitz
@bdr1177
Sep 06 2017 15:24
Also, I think the answer to your question about the prediction lines going low with later breakfast is yes. When the carb ratio changes, the OpenAPS prediction will change. Our breakfast CR of 1:17 used to change over to 1:45 at 9 am. If our daughter ate breakfast close to that time, the predictions would suddenly show her going low once the ratios changed over at 9 am.
jcwarrior
@jcwarrior
Sep 06 2017 15:24
maybe we have to set lower breakfast CR on pump, although we use higher CR to estimate bolus
That's what I was thinking
we are already doing the calculations out of the pump because it does not support sensitivity of 600
Brian Rabinovitz
@bdr1177
Sep 06 2017 15:26
If I thought a CR of 1:17 was correct, I would just ignore the prediction. The problem is that after the bolus snooze, I think OpenAPS will start to take the lower prediction into account.
jcwarrior
@jcwarrior
Sep 06 2017 15:27
that's right. Most of the time I see that it establishes a basal to 0 although it does not really need it
Brian Rabinovitz
@bdr1177
Sep 06 2017 15:28
An alternative to changing the pump settings is to turn off OpenAPS for a period of time to prevent the zero temp.
jcwarrior
@jcwarrior
Sep 06 2017 15:28
is not very important because the programmed basal rate is 0.025 at that period, so it does not stop getting too much insulin
but turn off OpenAPS at school prevents protection against real lows
Brian Rabinovitz
@bdr1177
Sep 06 2017 15:31
True. If you're not having problems with the zero temp causing highs, probably don't need to worry about it.
jcwarrior
@jcwarrior
Sep 06 2017 15:31
i will try to change breakfast CR on pump configuration and observe the behavior of openaps after the change
Brian Rabinovitz
@bdr1177
Sep 06 2017 15:31
I didn't realize any of the loopable pumps could adjust 0.025 increments. Is that the Medtronic 723?
jcwarrior
@jcwarrior
Sep 06 2017 15:31
medtronic 754. The medtronic 723 also support it
thank you @scottleibrand @danamlewis and @bdr1177 for all ideas. Sure we will find the best setup :-)
Brian Rabinovitz
@bdr1177
Sep 06 2017 15:33
Good luck!
jcwarrior
@jcwarrior
Sep 06 2017 15:34
:-)
Dana Lewis
@danamlewis
Sep 06 2017 15:49
:+1:
Craig Brenner
@cbrenner
Sep 06 2017 16:45

Well, I’m at my wits end. For the life of me, I cannot figure out how to tune the OpenAPS to handle the extreme aftereffects of very high fat meals on my son’s BG levels.

Previous to the OpenAPS, we know that almost exactly 2 hours after eating a high fat meal, my son’s BG would steadily rise upward to a 300 – 400 BG if we didn’t put in a an 8 hour/150% temp basal (immediately after finishing a high fat meal). If we missed putting on a temp basal after the meal, he would become super insulin resistant and we would have to give 2x corrections several times to get him back down until the 150% temp basal kicked in. Generally, if we got the 150% temp rate on right after the meal, this would work well and keep him in range for the ensuing 8 hours. Unfortunately, with the OpenAPS, it reacts too slowly and gets behind the 8-ball and 2 hours after a high fat meal, his BG goes through the roof (see below picture and similar rise the previous day in the 24 hours view). The OpenAPS does kick in with maximum temp basals when the BG rises but it is too little/too late and we end with very high BGs for the next 3-5 hours. We have to give manual corrections to help things along which subsequently causes the OpenAPS to counter with a lower temp rate (which is exactly what we don’t want) and things ultimately see saw for the next several hours.

Per Dana’s @danamlewis suggestion, I changed the pump’s max basal rate to 4.5 and tried changing the “current_basal_safety_multiplier" from 4 to 4.5 and to 5. And set the “max_iob to 4 and 4.5”. I have also set temp targets to 80 before and after these high fat meals to help things along. Unfortunately, none of this has worked well. On the contrary, these changes have actually caused considerably more hypoglycemic events during other times and have done little to smooth out the initial quick rising BGs 2 hours post high fat meals.

It seems to me that a prerequisite to deal with these high fat meals is that we need to have some type of temp basal rate kicking in right after the meal. So, 2 hours later, the extra insulin needed for the increased insulin resistance (or a much lower ISF), is ready and waiting and stops the rise. Thereafter, the system will need to maintain a higher temp basal for several more hours to deal with the impact of the triglycerides/fat on his BG. Perhaps a switch that can be set to indicate a high fat meal has been consumed so the OpenAPS knows the person’s insulin resistance is very high for the next 6-8 hours.

Any and all help with solving this maddening dilemma would be greatly appreciated!

High Fat Picture.JPG
Dana Lewis
@danamlewis
Sep 06 2017 16:47
Many people who are they sensitive enter some fats/proteins as a carb equivalent
Ie 5g carbs for every x of fat/protein
That would give you more bolus up front, which is what it sounds like you need given initial rise?
cameronrenwick
@cameronrenwick
Sep 06 2017 17:02
is it dev 0.5.3 that has the fiasp curve incorporated into it? Just on 0.5.2 right now but am planning on updating
Scott Leibrand
@scottleibrand
Sep 06 2017 17:03
no, exponential curves will be in 0.6.0
but please do update to 0.5.3 dev and comment the PR with your results
cameronrenwick
@cameronrenwick
Sep 06 2017 17:06
thx @scottleibrand .. will do so tonight and PR..
Craig Brenner
@cbrenner
Sep 06 2017 17:06
@danamlewis I know we tried this in the past and it was pretty tricky as he would go really low sometimes and other times it wouldn't be enough. The challenge here is counting fat grams --which is a bit harder than carbs when you are not at home and w/o nutritional labels. If you look at his bolus for this meal, he had 150g of real carbs and the fat was probably in the neighborhood of 80g (maybe a bit more or less but it was restaurant food). So if you did like 50% equivalent , he would be having 40g carb equivalent and that would put him at almost a 200g bolus (15u insulin). Just very concerning with that much bolus going in based on a lot of 'guesstimating' on both carbs and fat. Ultimately, we really want to reduce his T1D load and paying attention to fat for EVERY meal is just another burden we were hoping to avoid with the OpenAPS and other commercial closed loops down the road. The 150% temp basal rate for 8 hours has worked well w/o having to do much thinking on fat grams and carb equivalents..... so really prefer not to have to do carb equivalents...
Dana Lewis
@danamlewis
Sep 06 2017 17:08
@Kdisimone may can chime in, she's more in tune with what people do for equivalents (although 50% seems high for me). It may be easier to add 1-2u (or whatever number you decide is beset) to a fatty/protein dinner, to help deal with that up front. or, what you could do is set your 150% temp basal for 8 hours if you think that's best. but that's a double edge sword; the % basal type will prevent low temping if that's what's needed, but it would get you that 8 hours of doing what you wanted it to do.
(and at the end of hte % basal type you'd need to switch back to u/hr to be able to loop again)
Craig Brenner
@cbrenner
Sep 06 2017 17:12
@danamlewis So nothing can be added to the OpenAPS algorithm and preference settings to address people that are super sensitive to fat / protein?
Dana Lewis
@danamlewis
Sep 06 2017 17:14
I'd suggest trying some of the strategies that have been working well for everyone else, first. I personally don't do fat/protein equivalents, so I really would get some input from others on different strategies to do that. But the issue is you're not getting enough bolus up front, given the rise in the first hour-two hours time period. you could increase basal then, too, but that's not going to help as it's more reactive. so we need to figure out how for you to get more up front, which is going to be likely to be an equivalent (or a rule of adding x units to a super fatty/protein meal) to get less of an upfront rise.
also, are you running autotune? what does it think about your basic carb ratio?
Craig Brenner
@cbrenner
Sep 06 2017 17:18
@danamlewis Yes, running autotune automatically each night. Carb ratios (only pulls the a single one from midnight) is pretty close (if anything, it is suggesting we are a tad aggressive).
Dana Lewis
@danamlewis
Sep 06 2017 17:23
k
also, have you noticed what netIOB has been going into these meals?
Craig Brenner
@cbrenner
Sep 06 2017 17:26
@danamlewis BTW, I am also concerned with adding the extra bolus upfront for the fat as the OpenAPS sees all this extra IOB and predicts a low BG for a sustained period and keeps the temp rate at 0 much too long. This is actually counter to what is needed to keep the BG from continuing to rise over the next 4-6 hours. We actually don't want a 0 temp rate in these cases. In regard to netIOB, I will have to look at the logs but he is doing "Eating Soon" an hour or more before he eats.
Dana Lewis
@danamlewis
Sep 06 2017 17:30
That's why entering an equivalent in helps so it has that on the prediction
Craig Brenner
@cbrenner
Sep 06 2017 17:59
@danimaniac Thanks for the idea. In order to remotely login to the rig, though, i am pretty sure you would have to run a TeamViewer process on the rig.
Scott Leibrand
@scottleibrand
Sep 06 2017 18:02
@cbrenner why are you assuming that carb entry = up front bolus? I don't think that's what Dana is suggesting
if you bolus the same amount, but enter more carb equivalents for the fat/protein, then AMA can high temp to cover those as needed
and when it's not high temping, the extra carbs will prevent it from zero temping
and for teamviewer to work, you'd need to log into another machine on the same network as the rig and ssh from there to the rig. if the rig is mobile, you'd need to set up openvpn or reverse ssh tunneling
Craig Brenner
@cbrenner
Sep 06 2017 18:09

@scottleibrand Re-reading what Dana said, I do think she was suggesting both adding carbs AND also a bolus,

"...That would give you more bolus up front, which is what it sounds like you need given initial rise? and "It may be easier to add 1-2u (or whatever number you decide is beset) to a fatty/protein dinner"

That said, i am wondering if your idea of just adding the equivalent carbs would work and would be enough where the AMA could actually counter the rise with high temping. It would seem to be the equivalent to setting a manual temp rate, yes?

@scottleibrand As to TeamViewer. This is to connect to my son's rig and, other than at home, I would probably never have a machine on the same network at his location and (as we discussed earlier) the OpenVPN or reverse ssh tunneling is not for the common techie :smile:
Scott Leibrand
@scottleibrand
Sep 06 2017 18:20
If you're sufficiently motivated you could certainly do it. there's just nothing documented yet.
and yes, entering more carbs and having AMA high temp is similar to setting a manual high-temp, but more responsive to actual BG conditions
cameronrenwick
@cameronrenwick
Sep 06 2017 18:27
did 0.5.3 get hidden somewhere? found 0.5.2 on master and 0.6.0 on dev but no .3 ... what am I missing?
Dana Lewis
@danamlewis
Sep 06 2017 18:28
0.5.3 is the dev branch right now
cameronrenwick
@cameronrenwick
Sep 06 2017 18:28
oy...ok thx!
dan lightner
@danimaniac
Sep 06 2017 18:36
@cbrenner Yes, you would have to run an instance on the rig. Our rigs (assuming Edison/EB or Pi) are simply Ubuntu based linux servers which would likely accommodate a TeamViewer host module: https://www.teamviewer.com/en/download/linux/ There is also a module specifically for Pi. Again, just thinking out loud.
Paul Andrel
@stavlor
Sep 06 2017 18:37
that would likely need an X.org Environment to be capable of running on a rig
Yet alone the fact that teamviewer uses wine cringe!
Scott Leibrand
@scottleibrand
Sep 06 2017 18:55
Yeah, teamviewer is a GUI thing. You would want a CLI based solution for direct rig access.
Craig Brenner
@cbrenner
Sep 06 2017 18:58

@scottleibrand Ok, we will try entering ONLY a carb equivalent for the fat (i.e., no bolus) along with the normal carbs and bolus and see how that works at smoothing the initial rise.

It would still be great if this could be automatically managed by the OpenAPS so we don't have to be thinking about the fat in EVERY meal. As I said, the goal with moving him to a closed loop is to reduce the load. Wondering how the other closed loop systems are handling the rise and sustained insulin resistance due to high fat / protein meals.

Dana Lewis
@danamlewis
Sep 06 2017 18:59
They're all pretty similar - if told about carbs, they give more insulin as needed, otherwise they're responding to the resulting rise in BGs
(And given the timing of insulin, that's a much slower impact than if told about carbs/food up front)
Craig Brenner
@cbrenner
Sep 06 2017 19:02
@danamlewis Ugh! :cry:
Cameron Chunn
@applehat
Sep 06 2017 19:03
@cbrenner i feel that pain. I just can't eat pizza. I will rise and stay high for hours.
Carbs + fat just reck me.
David
@davidkeddydb
Sep 06 2017 19:04
if testers are needed for the more aggressive smb I would test it.
Dana Lewis
@danamlewis
Sep 06 2017 19:04
It's a YDMV, thing - some people need carb equivalents for protein fat, some of us don't. But probably takes some experimenting to figure out what strategy works best for calculating/entering them
Craig Brenner
@cbrenner
Sep 06 2017 19:05
really would like to get out of the fat counting! @applehat Have you tried what @scottleibrand is recommending with adding only carb equivalents for fat upfront on top of your normal carbs and bolus?
Phil
@drnoname82_twitter
Sep 06 2017 19:05
We have been seeing an awful lot of different pump-rig communication errors in the past two days, sometimes for hours on two different rigs (No acknowledgement from pump on wakeup, 'Pump' object has no attribute 'model, Error decoding FourBySix packet etc.) We hardly ever had radio issues before and mmtune results are still usually between -50 and -70db. Once radio communication is established, it is again solid with the usual good range. Pump battery is always over 1.5V.
Is there any known explanation for this behaviour?
Cameron Chunn
@applehat
Sep 06 2017 19:09
@cbrenner No, but I also follow a "be as lazy as possible" method. I have found that if I can get the initial bolus down just right carb-wise, my spike is manageable (maybe 180-190).
My goal with my OpenAPS rig is to not have to do any extra work in management. I literally just use my pumps bolus wizard and tell it when im eating something, and 95% of the time the outcome is perfect. I just learn to either avoid or live with the 5% cases. heh
I don't even pre-bolus for most meals
Unless I think about it before hand.
Im also an adult tho. I know its a lot harder for kids.
David
@davidkeddydb
Sep 06 2017 19:31
If I want to test what is currently up in 0.6.0-dev, do I just run cd ~/src/oref0 && git checkout 0.6.0-dev && git pull. Sorry I am not that git savvy
katie disimone
@Kdisimone
Sep 06 2017 19:44

@cbrenner we (my daughter is 15yo with t1) do have to manage proteins and fats. She definitely can't eat just even plain chicken without needing insulin. Prior to looping, we tried to find a ratio based on protein type and grams of protein/fats. It's just never worked out well. Way too complex and I just wasn't interested in that much work when the results still came out varied for other reasons (that seems to be the diabetes way). So, instead...we just sort of learned over time how much to eyeball equivalent carbs of certain things...a hot dog is about 5g, each egg is 3g, a heap of shredded chicken is 6g, a chunk of red meat about 8g.

It's not super tight estimates as a result of eyeballing things, but overall, it's worked well over time. The conversations usually go like this now:

Me: how much does the package say?
Anna: 35g grams, but I'll probably round up to 40g

Her rounding usually is just based on enough experience with similar types of food in the past or past spikes from same food.

Things get a little complicated when fats are significant though because all the behavior changes for when we can bolus for those extra grams. So she needed help on OpenAPS to know how to split those total carbs up best (we prebolus 20 minutes for first part of bolus for a meal...usually did second part of split after finishing the meal). We found that once we hit about 4-5 hours though with meals...if carb absorption was still going, we'd need to enter in more carbs to help OpenAPS realize more absorption was still happening (aka the algorithm had decayed the carbs a little too quickly) to help the system keep control.

David- I think you'd want to git pull first to bring in the 0.6.0-dev branch and then check it out.
David
@davidkeddydb
Sep 06 2017 19:51
Thanks Katie. I am still reading all the docs that have been posted.
cameronrenwick
@cameronrenwick
Sep 06 2017 19:57
@cbrenner I'll chime in here with similar findings as @Kdisimone has indicated. I'm the T1 so if I over/under estimate.. it's me who gets the knock. The one thing I've found (and not mentioned here) is about activity post meal. My job often is pretty active with lots of post meal activity but, other times, I'll be at a computer entering notes for an hour. This activity (or inactivity) also has to be taken into account (in my case anyway) and adds an extra layer of YDMV to the mix. Not easy for sure to bring all these elements together. Anyway, long story short, I try to consider what my day will entail when considering food and particularly how carby/fatty/lean my diet is with activity level throughout the day. If I'm going to have a busier day, I'll get away with fattier/carbier meals... if I'm more sedentary, I try to eat really well. Every day is learning but without a doubt, looping, smb, uam, and fiasp are making things easier. Dana and Scott are super-infosharing heros :sparkles: :heart:
Scott Leibrand
@scottleibrand
Sep 06 2017 20:11
David and @Kdisimone it's usually best to git pull both before and after a checkout. The latter is for if you've already checked it out once before and want any updates since.
katie disimone
@Kdisimone
Sep 06 2017 20:13
@scottleibrand yup...I was remembering that time with decocare where I'd checked out first and had problems.
Scott Leibrand
@scottleibrand
Sep 06 2017 20:13
Technically you only need fetch before checkout, but pull does a fetch, so that's fine too.
David
@davidkeddydb
Sep 06 2017 20:15
@scottleibrand @Kdisimone so that does that look like cd ~/src/oref0 && git pull && git checkout 0.6.0-dev && git pull or did I completely screw that up?
katie disimone
@Kdisimone
Sep 06 2017 20:15
That would work
Scott Leibrand
@scottleibrand
Sep 06 2017 20:16
:+1:
Then you'll need to npm run global-install too
And probably re-run oref0-setup from oref0-runagain.
JELCRAWFORD
@JELCRAWFORD
Sep 06 2017 20:17
@Kdisimone thanks for the info about your split bolusing and other considerations. It is helpful to have ideas about how other parents manage a variety of presentations. The permutations of things to consider seem practically infinite :)
Scott Leibrand
@scottleibrand
Sep 06 2017 20:18
You can also move your myopenaps directory if you want to create a new one with --nogit