We ran into some of what you’re describing and altering the carb ratio seemed to do the trick for us. Our basal tests were coming back fine, ISF is harder to nail down for us but we were close, yet we were getting stuck high in the afternoons. Eventually, I traced it back to a bad breakfast carb ratio. She was getting too little insulin at breakfast and morning snack, and the numbers were slowly creeping up. Then she’d eat lunch, her AM vs. PM ISF would kick in, and all that would catch up to us resulting in high numbers hours after breakfast was out of her system. Once we tightened up our breakfast carb ratios, things got a lot easier in the afternoons.
7 Likes
UPDATE: The Inadvertent Basal Test
So, Samson had to get a follow-up MRI yesterday to see if the autoimmune encephalitis he was diagnosed with in October has completely resolved. (We suspect it has as he hasn’t had a seizure since discharge, but please keep him in your thoughts and prayers as it’s scary until we get the results). Because he’s little he had to be on anesthesia. Which means he was NPO for 12 hours prior to the scan and no fluids for 2 hours prior to the procedure. AND…that meant it was a very convenient time to a do a basal test. Our endo said he needed to be Looping beforehand, to reduce the risk that he’d have lows prior to the anesthesia, so we couldn’t do a perfect basal test, but here’s a basal-test-with-Loop-running from two nights ago.
It’s tough to generalize, as he had pizza, bell pepper and mochi for dinner the night before, so he actually had quite a few carbs and fat, but on first glance it seemed like the nighttime basal rates was roughly appropriate. The ISF also looked good overnight, as he was high and Loop gently brought him back down to target before the morning. However, he did trend low, even with Loop running, in the early morning hours. Then, he slowly rose from about 9 a.m. to 11:15 a.m.; he was 183 mg/DL when we disconnected his CGM for his MRI. At 12:06pm he was 118 mg/DL on finger stick, after being off his pump for that period of time, and at 2:20 p.m. he dipped to 63 mg/DL. The MRI and anesthesia makes the immediate aftermath sort of useless for generalizing because who knows what was going on in his body to cause that drop. But it does makes me wonder if his afternoon basal rate isn’t too high.
Then, overnight tonight, the RileyLink wasn’t running for some reason. I slept through many many low alarms 
. And he was very low over night. I feel awful. However, there do seem to be step-wise drops in his BG that correspond to basal changes in the early morning hours starting at around 3am. So that makes me think that we definitely need to drop his morning basal rates. But his 12am to 3am rates seemed pretty steady so for now I’m going to keep them as-is.
So his new settings:
Basal rates (units/hr):
12:00 a.m. | 0.125
3:00 a.m. | 0.175
4:00 a.m. | 0.2
6:00 a.m. | 0.225
8:00 a.m. | 0.225
12:00 p.m. | 0.4
2:00 p.m. | 0.225
7:00 p.m. | 0.275
9:00 p.m. | 0.4
10:00 p.m. | 0.425
11:00 p.m. | 0.3
4 Likes
There are a few things I do when this happens. First I put the RL quite close to the pump just to rule out radio interference. I go into loop > settings > pump > devices (rileylink) > scroll down and “send button press”. If that succeeds, then everything’s working again. If it gives an error message, then I reboot the RL by turning off that tiny recessed switch in the side, and after a moment I turn it back on on and watch for the RL lights to activate so I know the RL is operational. I keep a toothpick in the glucometer case so that I always have a small-enough tool to operate the RL switch. After rebooting the RL this way, I go back and try the “send button press” again and it pretty much always works. If it still doesn’t work, then I reboot the phone and restart the loop and dexcom apps.
So you would want to lower the basal starting around 1am or 1:30, because it takes time for the change to have effect.
3 Likes
Our RileyLink is really on its deathbed and we have a new one on backorder – it basically only works when it’s plugged in, and it has to be plugged in just-so in order to charge. I keep seeing this blue-light emanate through the case when it’s not working, and a green and red light seem to be the indication that ti is charging.
As an aside, The reliance on an external computer is my #1 complaint about Looping. I know that we’d have much better results if we knew RileyLink wouldn’t just crap out or lose connection randomly once it got older…
2 Likes
I’m not sure it takes that long for the basal changes to kick in; Medtronic pumps give four or six pulses of basal an hour, which means by 30 minutes in, you’re seeing an increase in insulin. I mean I see three step-wise drops about 45 minutes after a new basal rate was set; he’s dropping faster after each incremental change, about an hour after each change.
But if that doesn’t do the trick I can always change the earlier basal. I’ll see tonight.
2 Likes
Standard medtronic pump, not looping, will vary the number of pulses per hour, to match basal rate for that time period. Could be more or less than 4 pulses per hour. Each pulse gives same amount.
How does Loop work ? Do you set physical pump basal to 0, and Loop calculates and delivers pulses (mini-bolus) as basal. And in your case results in 4 per hour?
2 Likes
your pump still has basal rates programmed in; if Loop is not running, it defaults to those programmed pump basals. So there’s never a situation where your physical pump basals are set to 0.
It’s always been a little bit mysterious to me how those incremental changes in basal (Loop might run every minute or two) are actually translated into miniscule changes in insulin delivery via basal, if they are at all. I wish I understood that better, honestly. My son’s pump is limited to increments of 0.025 units per hour; I don’t know if the basal at that low rate is 0.025 units in one pulse, or in four, but it’s honestly baffling me to how, say, a basal rate of 0.375 units/hr, turned on for 3 minutes, then cancelled and switched to a 0.2 unit/hr basal rate, actually translates into insulin delivery.
2 Likes
One thing you can do on minimed is check under Utilities - Daily Totals. It will show basal so far that day, then check every 5 min to observe how delivery accumulates total.
Info here might be helpful.
If your setting is 1.9 u/hr, the pump will deliver in 0.05 unit amounts. 1.9 units divided by 0.05 doses is 38 deliveries over one hour, or 0.05 units every 1 minute and 35 seconds.
For 2.9 u/hr, the pump will still deliver in 0.05 unit amounts. 2.9 units divided by 0.05 is 58 deliveries required over the hour, or 0.05 units every 1 minute and 2 seconds.
For people with basal rates under 1.0 u/hr, the pump delivers in smaller doses of 0.025 units. For example a setting of 0.6 u/hr divided by 0.025 is 24 deliveries over the hour or 0.025 units every 2 minutes and 30 seconds.
3 Likes
I wanted to know with Omnipod. I just left it running for 3 basal periods (0.2u, 0.6u, 1.2u I think) and observed moving the pump along the paper after each pulse and noting time
I am not sure how practical it is with Medtronic and your pump change schedule
Omnipod pulses the minimum (05.u) up to 12 times per hour.
If more is needed i.e. basal >0.6 then it changes accordingly within those constraints and to allow even distribution.
e.g. 0.7 basal is 0.1 x7 times per hour etc.
you may be able to assess the insulin stain to guess the pulse size - I didn’t think of that. I suppose circle it with a pen in case it dries out
Ed
5 Likes
I think for Samson at night, he’s getting between five and eight pulses then, since his basal rate is between 0.125 and 0.20 right now. I guess during the day it could be up to 16 pulses.
3 Likes
Sounds good. When rate change at time X, pump just determines new “wait” time between pulses.
1 Like
maybe - it may depend how long it takes to complete a max delivery of basal insulin I wouldn’t be surprised if 12 was the max per hour- openAPS logs with Medtronic seem to be every 5mins also.
2 Likes
I’m confused by your term “delivery via basal”. And intrigued!
My understanding is that OpenAPS can only communicate to Medtronic pump as a “remote” bolus command, although may be referring to it as a basal change. If that is true, a basal bump from APS would be timed per APS logic, maybe a .025 “bolus command” every 10 min until it sees BG starting to come down. But standard pump basal continues on its pattern/timing, not considering that as basal.
It would be interesting to compare stats, if OpenAPS stores history of these “micro boluses”, and ideally if it was triggered as true meal bolus vs a correction/basal squirt.
So if you were looking at pump history of boluses, would you expect to see basal bumps as bolus? Or do you just rely on OpenAPS data?
2 Likes
Basically, Loop sends a command to the pump to issue a half-hour-long temporary basal of a certain amount. It runs its algorithm every minute or so, and if it decides the currently assigned temp basal is too high or low, it then sends a second command to the pump to cancel the temp basal and issues a new temp basal. That’s how it ramps delivery up and down. The idea being that if you get away from the RileyLink or are unable to Loop, the temp basal shuts off and you’re left with the regularly programmed basal rates. Since temp basals override the programmed basals while active, it’s as if you’re just constantly running on different temp basals.
I think the way openAPS and Loop keep track of this delivery is via the volume left in the reservoir. Adding up all the half-finished basal rates would get confusing and is very error prone.
As a side note, because these commands are issued as temp basals, you will not see them registered as IOB on the pump itself; you need to look on NightScout or on the Loop app, etc., to see how much IOB was given in the form of elevated temp basals.
Does that make sense?
4 Likes
Probably splitting hairs, but I think this means Loop sends a series of commands (micro bolus, at the right time), to emulate as though a temp basal was manually programmed into pump.
If you compare on-screen pump total bolus and basal history (by day) to Loop, I’m curious how it would look. On pump screen, it is under Utilities - Daily Totals.
2 Likes
Not that it matters much, Loop runs its algorithm every 5 minutes, timed by a clock on RileyLink and initiated by a Bluetooth wake-up call from RileyLink to iPhone. On medtronic pumps, a new temp basal command overrides any already running temp basal, so there is no need to cancel it - a single temp basal command is sufficient. On omnipod, I think they must first cancel the running temp basal and then send a new temp basal command.
Not exactly. Both medtronic and omnipod have temp basal commands available, and both Loop and OpenAPS make use of the temp basal commands, there is no need to emulate that behavior. The pump itself executes such commands in the form of a series of 0.05U increments (in some cases 0.025U increments, depending on the basal rate and pump capability). Your description above of how basals actually work is correct.
4 Likes
There is already tons of good feedback and updates in the thread above. I would agree that it may take surprisingly long time to digest certain food (such as cheese or beef in my case, up to around 8 hours, sometimes even longer), which makes it difficult to evaluate basal rates for a human or for an algorithm. For example, Autotuning (from OpenAPS) makes certain assumptions about food absorption, which of course affects its recommendations for parameter settings.
Related to the above, and related to parameter settings, I personally find the Loop’s “Carbohydrates” screen extremely useful. In the Glucose Change graph, Loop displays the “insulin counteraction effect” (ICE), which is the difference between the observed 5-min glucose change and the glucose change expected based on the insulin enacted over the same 5min interval. For example, lets say that glucose dropped by 5 mg/dL over 5 minutes, and that the modeled insulin effect was -10 mg/dL over the same 5 minutes. The ICE would be (-5) - (-10) = +5 mg/dL over that 5-min interval, which would be displayed on the ICE graph as a (+5/5 min) = +1 mg/dL per minute. This would imply that something (typically food absorption) is pushing bg up at the estimated rate of +1 mg/dL/min. In the absence of any food absorption, a positive ICE implies too low basal rate or too high ISF setting. If there is lots of IOB at the time, the most likely issue is with ISF. If there is little IOB at the time, the most likely culprit is the basal. For all of this to work out, the insulin absorption model must be close to reality (which is why people should not use the Walsh model in Loop).
The same approach is used by Loop to make estimates of actually absorbed carbs compared to what was entered, and the results are also displayed on that same screen. A lot can be learned from these results, including how good parameter settings are, and also how certain food actually absorbs over time.
4 Likes
Erin and I continue trying to dial Liam in also before school. At this point, I feel I’m Ill-equipped to “pass off” our looping to anyone else at school or wherever. For us… Patterns that we see and adjust for go the other way after the adjustment has been made. For boys as small as ours .05 really is a huge deal sometimes. I am having the problem right now of Liam running low overnight in closed loop (so no basal being administered for long periods of time) and giving him unbolused snacks that do bring him up, don’t keep him up. I can’t think of what it could be if not for still honeymooning.
Lots of challenges in our household also Tia. Have you thought of the possibility of going low carb JUST FOR SCHOOL? I’ve been giving that more and more consideration lately because the level of interaction and effort I put forth now I can’t expect anyone other than me to ever be able to keep up with it… It’s non stop really.
Loop helps, but it’s still a lot of work.
2 Likes
@ClaudenDaye, LOL I mean, the idea of going low carb is great but unfortunately Samson eats no meat, no eggs, fish only in very very limited circumstances and basically no protein except a handful of beans and cheese. I mean his control would be great because he basically would not be eating anything all day. Or it would be abysmal because he’d be sneaking the horrid school lunch.
Also his school offers two “snacks” (they are anti-low-carb things like rice krispie treats or other junk that parents bring in). I guess it’s for kids who are on free- or reduced-price lunch; some of them don’t get enough food to eat otherwise. But it’s been the bane of our school experience for our older son, whose food tastes have degenerated as a result of having those things every day.
Anyways, even though I despise those snacks, I don’t want to single Samson out and say he’s the only one who can’t eat it.
If it helps you any, Samson has roughly the same sorts of numbers as Liam I suspect, and he’s been in preschool for a while, with crummy diabetes treatment and teachers who are not even trained to use a bolus wizard, let alone carb count. The microtweaking and etc. that I do at home is just not possible, and in the end, it sometimes works out worse, sometimes better, but overall it’s okay. I think if you work hard at dialing in some parameters and stay on top of updating them when school comes, I think it’s very possible to get a decent level of control with, say, one bolus for lunch and snack, and instructions to bolus when he’s high over 200 or something.
Also, I don’t know how Erin and Liam do things at home, but we find that Samson has a much more routine eating pattern at school. He eats basically the same lunch every single day, at the exact same time. So that can help you figure out a program that might work decently well.
3 Likes
This sounds really really helpful @dm61. I think I need to reread this later and then take a look when Samson’s phone is nearby. I think it could help us with fine tuning.
4 Likes