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

21st
Feb 2015
Scott Leibrand
@scottleibrand
Feb 21 2015 00:37
And... there ya go! (Got started on it during an All Hands call, and got far enough along that I decided to just finish up the first draft.) :-)
LMK if you see any glaring errors, and then it's probably worth sending out to the openaps-dev list.
Dana Lewis
@danamlewis
Feb 21 2015 00:41
I want to send it out along w the spreadsheet of who's doing what so ppl review both.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 15:56
I sent Dana and Scott the requested article. Basically, you would announce the exercise in your OpenAPS and let the 6 formulas account for the exercise. It looks very simple to apply. There are 2 derivatives, I believe, which is the hardest concept to apply.
I haven't read the article though yet. Remember that reading isn't my thing!
so, @TC2013 , when you were talking about the "length of the hypotenuse" I am pretty sure that is equivalent to magnitude
Toby Canning
@TC2013
Feb 21 2015 19:25
@diabeticgonewild Can offer an insight into Duration of Insulin Action? I'm trying to put this into perspective: http://www.diabetesnet.com/diabetes-technology/insulin-pumps/pump-insulin-duration
diabeticgonewild
@diabeticgonewild
Feb 21 2015 19:46
Will take a look at it, right now. Getting on the computer that has the screen reader, Kurzweil 3000. I also have John Walsh's book, pumping insulin. The DIA can be modeled by a natural logarithm (least ideal), 2 2nd order Linear Regression equations , or differential equations (most ideal) and possibly even more ways
diabeticgonewild
@diabeticgonewild
Feb 21 2015 19:51

Actually, this is a very good thread on CWD to read that simplifies it. I found that the Animas pump, at least for me, had the most accurate representation of DIA. http://forums.childrenwithdiabetes.com/showthread.php?15364-Extrapolating-curvelinear-IOB-charts-from-Animas-and-Paradigm

Actually, I do know twodoor2, and her husband is an EE. Maybe I can get them in on the OpenAPS project.

twodoor2 is a mathematician
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:08
OK, this part isn't insight but: 1. Avoid stacking boluses for any reason
  1. Conservatively (err on the side of a longer DIA) setting to avoid lows from DIA calculuations, and then subsequent highs from the lows.
  2. (Potential, but controversial insight, especially for CWD) Consider (with doctor approval) Afrezza as it has short DIA, reserved for resistant highs only (I might have to, as I might be on long-term glucocorticosteroids)
The Afrezza is risky and adds more variables, though, but it doesn't sound as bad as persistent highs to me, with all I have going on.
An interesting part about what he had to say about children (I was looking some of this up too):

From here (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3038395/) :

  1. Children have nearly three times the body surface area (BSA) to body mass ratio of adults.
  2. Children younger than two years have thinner layers of skin and insulating subcutaneous tissue than older children and adults.

From here:

  1. When insulin is delivered in the form of a large drop, the surface-area-to-volume ratio is quite low, so the diffusion rate is relatively slow. But if the insulin is delivered in the form of thousands of tiny droplets, the surface-area-to-volume ratio increases, and the molecules diffuse much more rapidly. - See more at: http://diatribe.org/issues/39/thinking-like-a-pancreas#sthash.43jbYeab.dpuf
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:13
So, in CWD, depending on the age, it can be postulated that when the surface area to volume ratio is lower, the DIA is longer. It can also be postulated that it would be more variable in children (not because of the fact that they are growing) but due to having thinner layers of skin and insulating subcutaneous tissue.
*even longer than adults
Scott Leibrand
@scottleibrand
Feb 21 2015 20:18
Bolus stacking is not something to avoid, as long as you (or your pump) calculate IOB properly.
6U of insulin delivered as four stacked boluses over 2h is way safer than 6U all at once, because the peak activity is spread out by 2h.
Ben West
@bewest
Feb 21 2015 20:19
yeah, the literature's advice against stacking is unwise
Scott Leibrand
@scottleibrand
Feb 21 2015 20:19
So you're much better able to counteract with carbs if needed.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:19
I guess, but I am one of those people who is a fan of eliminating variables. Oh, I forgot, apply correction basals or "superboluses". That's a good rec.
Ben West
@bewest
Feb 21 2015 20:20
looking at the people with the best performing glycemia, they are stacking 8 doses deep
Scott Leibrand
@scottleibrand
Feb 21 2015 20:20
The advice against stacking is for people who don't track IOB.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:21
Like IDK, I am a pretty compulsive person, and I have my "personal rules". I don't rely on IOB actually, cause I am really good at math. But that's just arrogance, too.
Thanks for the correction. I appreciate it.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:26
Rather than model like the subcutaneous compartment and the plasma [blood
] compartment and everything else @TC2013 , the easiest approach for you would likely be do it the CWD thread way. I have an Animas 2020 pump if you want me to do the DIA for that pump (and others)
Ben West
@bewest
Feb 21 2015 20:27
if we avoid stacking, we can't do eating soon mode for example
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:28
Not necessarily. It depends on how you calculate or split the dosage, actually
Ben West
@bewest
Feb 21 2015 20:28
in eating soon mode, we are relying on bolus for food to be stacked against a correction to low end of target... the ongoing kinetics help avoid the spike
Scott Leibrand
@scottleibrand
Feb 21 2015 20:28
@TC2013: what exactly is the DIA question?
split bolus = stacked bolus
Ben West
@bewest
Feb 21 2015 20:29
stacking insulin is usually any time you give insulin while there is still previous insulin dose active
any dose, for any reason
Scott Leibrand
@scottleibrand
Feb 21 2015 20:29
exactly
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:30
I don't want to make it into a debate, and yes, I don't believe in stacking, but if you know you are going to apply a correction at X time and a meal with Y carbs at time T, I don't consider that stacking.
Ben West
@bewest
Feb 21 2015 20:30
that is the definition of stacking in the literature
Scott Leibrand
@scottleibrand
Feb 21 2015 20:30
Ok, let's not debate definitions. What kind of stacking do you consider unwise?
what would be the problematic scenario?
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:31
Like multiple meals or excessive "snacking" in a short period of time. I believe in spreading it out reasonably. I also believe in limiting CHO (to a certain degree--whatever is appropriate for the person--not a debate)
Scott Leibrand
@scottleibrand
Feb 21 2015 20:32
CHO = COB?
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:32
Like, not enough time for the BG to recover from the "meal/snack" excursion
I guess, COB.
Scott Leibrand
@scottleibrand
Feb 21 2015 20:34
or CHO=simplified C6H12O6 formula?
so in our experience, avoiding snack stacking is mostly important because people don't have a good model for carb absorption.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:35
I meant CHO. I don't use COB, mostly because I have severe gastroparesis (not diabetes-related) and so I just am not a fan of that concept. Sometimes my stomach empties normally and sometimes it dumps.
slowly
That's interesting. Well, no offense, but that isn't going to work for me if steroids are in my future, no matter what. I do believe you though. I don't like being a skeptic though.
Scott Leibrand
@scottleibrand
Feb 21 2015 20:36
Yeah, so if you can't reliably model COB, then limiting carb stacking is prudent.
Dana Lewis
@danamlewis
Feb 21 2015 20:37
Yea some ppl just don't know what cho stands for (aka carbs) good to clarify all around
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:37
Got it.
Scott Leibrand
@scottleibrand
Feb 21 2015 20:37
I don't think limiting insulin stacking is the issue though, except to the extent that you no longer need to do so if you avoid carb stacking.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:38
Agreed, come to think of it. Because correction dosages are needed no matter what.
Scott Leibrand
@scottleibrand
Feb 21 2015 20:38
insulin kinetics / pharmacodynamics are better understood than the variability in digestion for sure.
Ya
Toby Canning
@TC2013
Feb 21 2015 20:40
@scottleibrand You asked me what my question about DIA is. Well, we have been using 3 hour DIA settings for years. As I have been reading about how insulin actually works, even though we were getting good results, we must have been compensating with insulin sensitivity factors, etc.
In our autopilot mode, I am trying for fit a longer curve and I wanted a good way to do it.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:41
That CWD thread, as percentages. Easiest way.
Scott Leibrand
@scottleibrand
Feb 21 2015 20:41
I have a formula for 4h DIA in a github issue
Toby Canning
@TC2013
Feb 21 2015 20:41
I had found that earlier and wrote up the equations as such already
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:41
Don't worry about the subcutaneous absorption stuff, unless dosages vary from 0.5-25 units in a day.
*bolus dosages
Very nice @TC2013
I have an Animas 2020 if you want to play with it, for any reason.
Scott Leibrand
@scottleibrand
Feb 21 2015 20:42
my equations are two quadratic formulas, one from bolus to peak activity, and the other from peak to DIA.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:43
Yeah they are the ones that look like Fourier transforms
Toby Canning
@TC2013
Feb 21 2015 20:43
Right, I didn't know about the 4hr one, I only knew of your 3 hour.
Scott Leibrand
@scottleibrand
Feb 21 2015 20:43
@kenstack might have a better formula based on a nonlinear insulin activity curve
haven't gotten 4h into wip/iob-cob yet: it's still on the todo list
Toby Canning
@TC2013
Feb 21 2015 20:44
I was looking at that, definitely over my head
Scott Leibrand
@scottleibrand
Feb 21 2015 20:44
but the formulas are all there.
nightscout/cgm-remote-monitor#301
Toby Canning
@TC2013
Feb 21 2015 20:45
Is his adjustable?
Scott Leibrand
@scottleibrand
Feb 21 2015 20:45
i haven't used @kenstack's formulas if that's what you mean.
Toby Canning
@TC2013
Feb 21 2015 20:46
Thank you for the info!
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:46
All you have to do is do the procedure in that CWD thread, put the respective values into an Excel spreadsheet, and have Excel do a 2nd order least squares regression
Bam! Equation made
Toby Canning
@TC2013
Feb 21 2015 20:46
I know Ken said his follows the Medtronic curve really well
Scott Leibrand
@scottleibrand
Feb 21 2015 20:47
That excel method is very similar to what I did to get my formulas, except I started with insulin activity instead of CWD data.
Toby Canning
@TC2013
Feb 21 2015 20:47
We still use that manually and I'm trying to get the two systems to match up.
Scott Leibrand
@scottleibrand
Feb 21 2015 20:47
ours matched Medtronic's pretty closely as well.
But it's not directly derived from it.
Toby Canning
@TC2013
Feb 21 2015 20:49
Very good, now I have 2 or 3 options to test!
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:49
Animas was most reliable for me. Well right now I have a t:slim, Animas 2020, Medtronic 522, and Medtronic 530G in my closet, so I can try all of those or loan them
Scott Leibrand
@scottleibrand
Feb 21 2015 20:49
i modeled insulin activity as going linearly from zero to a peak at 60-90m, then dropping linearly back to zero at DIA
Toby Canning
@TC2013
Feb 21 2015 20:50
Why does 3 hours work for Dana?
Scott Leibrand
@scottleibrand
Feb 21 2015 20:50
The literature shows a curve that does something similar, and IIRC our approximation falls mostly within the error bars in the literature.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:50
MATLAB has curve fitting tools also, but unless you love programming, Excel is the way to go
Scott Leibrand
@scottleibrand
Feb 21 2015 20:50
4h would probably be better for closed loop
Toby Canning
@TC2013
Feb 21 2015 20:50
That doesn't seem to fit...
Scott Leibrand
@scottleibrand
Feb 21 2015 20:51
DIA actually varies a lot
Dana Lewis
@danamlewis
Feb 21 2015 20:51
I was 4 for a long time, watching the data and curves actually convinced me to bump back to 3
Scott Leibrand
@scottleibrand
Feb 21 2015 20:51
4h is more conservative, and better matches the situation where pump site blood flow is restricted and insulin pools up
but 3h seems a better match for daytime.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:52
The larger the dose, the longer the DIA. With kids the longer DIA makes more sense too, from what I posted above.
Toby Canning
@TC2013
Feb 21 2015 20:52
Yes, but every bit of literature I found showed 4 to 6 hour curves.... Is the tail just too tappered and therefore not significant enough to affect the results?
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:52
I have one DIA and it is set to 5 hours
Toby Canning
@TC2013
Feb 21 2015 20:52
Or longer curves for 10u +
Dana Lewis
@danamlewis
Feb 21 2015 20:52
(Given the attention to the system now, 3 hours works better and can always trigger resistance mode if I have pooling or something)
Yea
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:53
the tail is the amount of displacement. It is significant. Longer curves for more insulin. Due to surface area to volume ratio in particular
Scott Leibrand
@scottleibrand
Feb 21 2015 20:53
I think the thing that actually matters most is the peak activity time (60-90m).
Dana Lewis
@danamlewis
Feb 21 2015 20:53
I rarely ever do more than 3u at once, and even that not so much. Queen of stacking
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:53
I would agree with @scottleibrand
Scott Leibrand
@scottleibrand
Feb 21 2015 20:53
how long you stretch out the tail (DIA) doesn't matter as much, as most of the activity is around peak.
Toby Canning
@TC2013
Feb 21 2015 20:54
Roger that.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 20:54
correct, and you could even make predictions based on the first derivative test (where blood sugar reaches a peak from a meal and insulin kicks in to a point where the blood sugar doesn't change)
Scott Leibrand
@scottleibrand
Feb 21 2015 20:55
Ya. Wip/iob-cob makes that easy to visualize.
Toby Canning
@TC2013
Feb 21 2015 20:56
My son isn't as controlled as Dana, so I want the system to avoid overstacking. Thus I am making the DIA longer to help.
Dana Lewis
@danamlewis
Feb 21 2015 20:57
I didn't get to where I am without DIYPS.
Scott Leibrand
@scottleibrand
Feb 21 2015 20:58
Yeah, I definitely think we'll want to default to at least 4h DIA for OpenAPS.
Dana Lewis
@danamlewis
Feb 21 2015 20:58
See also all the data graphs that show my time in range and eag before DIYPS, aka just over a year ago
Scott Leibrand
@scottleibrand
Feb 21 2015 20:58
longer DIA just makes the system a bit more conservative.
Toby Canning
@TC2013
Feb 21 2015 20:58
right!
Dana Lewis
@danamlewis
Feb 21 2015 21:00
(All this talk of carbs made me want extra
gf toast ;))
diabeticgonewild
@diabeticgonewild
Feb 21 2015 21:00
This is the basic concept. In Excel it is called "best fit". https://www.youtube.com/watch?v=oP97cer687M You would want 2nd order (linear) regression, which is obviously a quadratic equation.
Sorry, didn't know that would happen
Scott Leibrand
@scottleibrand
Feb 21 2015 21:00
Another case where stacking is essential: when you have more than 60g carbs, blousing for it all at once would be dangerous. You need an extended/combo (stacked) bolus to spread the activity out to match the digestion.
Extra toast is a feature, not a bug. ;-)
diabeticgonewild
@diabeticgonewild
Feb 21 2015 21:01
True, but I still don't consider that stacking. But I guess that is my arrogance.
Scott Leibrand
@scottleibrand
Feb 21 2015 21:01
nah, just different definitions and assumptions.
Toby Canning
@TC2013
Feb 21 2015 21:02
Can DIYPS accomodate a dual wave/ square wave bolus?
Scott Leibrand
@scottleibrand
Feb 21 2015 21:02
That is often the case when people disagree: different assumptions make us talk past each other.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 21:02
True, I agree.
Scott Leibrand
@scottleibrand
Feb 21 2015 21:02
It did once upon a time. But we never used it, in favor of just doing half unit additional boluses as needed.
Just as effective as a square wave, but more adaptable as BG comes in higher or lower than predicted.
Toby Canning
@TC2013
Feb 21 2015 21:04
Our son would forgot to make the manual post-meal entries.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 21:04
I have never done a square wave bolus ever. I have always just done the math in my head and applied a bolus on top of a basal that lasted roughly until the peak of the bolus.
Scott Leibrand
@scottleibrand
Feb 21 2015 21:04
of course that only works when you have DIYPS telling you as soon as it's time to do another half.
Toby Canning
@TC2013
Feb 21 2015 21:05
even that wouldn't work for us...
Scott Leibrand
@scottleibrand
Feb 21 2015 21:05
Yeah, that's basically manual square wave. DIYPS closed loop does that if you don't bolus enough up front.
Toby Canning
@TC2013
Feb 21 2015 21:05
He ignores his pebble as it is.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 21:06
For good reasons! It's annoying!
hahaha
I guess you are going to have to account for that in the design, by making it less dynamic and more static, by applying super-boluses/square-waves/whatever, initially, and setting the target BG higher (or whatever else) to compensate
Scott Leibrand
@scottleibrand
Feb 21 2015 21:08
Our idea of how OpenAPS should work is you bolus normally for your meal, and then once those carbs hit, the system takes over and adjusts basals as needed to get you back in range.
Toby Canning
@TC2013
Feb 21 2015 21:08
Right... Now I need to leave and go figure this out.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 21:09
LOL, you will. Think "correction basals".
Scott Leibrand
@scottleibrand
Feb 21 2015 21:09
So for a meal up to 45 carbs, you just do a regular bolus for it. For anything more, you might square-wave.
Toby Canning
@TC2013
Feb 21 2015 21:09
Agreed
Scott Leibrand
@scottleibrand
Feb 21 2015 21:10
If BG drops too much, it cancels any extended boluses and low-temps.
If BG rises too much, it high-temps until you start coming back down.
Otherwise it waits to see how the BGs play out.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 21:11
which loops to IOB which probably loops to the target at time T.
Scott Leibrand
@scottleibrand
Feb 21 2015 21:11
we keep the target constant
diabeticgonewild
@diabeticgonewild
Feb 21 2015 21:12
yeah, of course. But the target is predicted to reach at time T.
I would think, right?
I believe IOB is the aggregate IOB of all basals (standard/correction/meal/whatever) plus boluses, combined, correct @danamlewis and @scottleibrand
Scott Leibrand
@scottleibrand
Feb 21 2015 21:13
Right.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 21:49

@TC2013 , for simplicity sake, I would go the CWD thread route, and just use a pump for referencing DIA information as percents. However, a more ideal way (although slightly more time consuming--possibly more imprecise) is looking at the package insert of the insulin of whatever your CWD is using. You would look at the graph for the absorption profile where the "peak" of the insulin represents 100% DIA and no remaining DIA as 0% . Using the graph and the values in between those extremes, you can interpolate the remaining DIA for the respective insulin he uses.

Apidra (see page 14) : http://www.accessdata.fda.gov/drugsatfda_docs/label/2014/021629s029lbl.pdf
Humalog (see page 8) : http://www.accessdata.fda.gov/drugsatfda_docs/label/2013/020563s115lbl.pdf
Novolog (see page 14) : http://www.accessdata.fda.gov/drugsatfda_docs/label/2015/020986s080lbl.pdf

Actually it would not be 100% duration of insulin action. It would be the maximum, 100% of concentration of insulin.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 22:00
and the percentages would represent how much of the insulin is in the serum [plasma] which is representative of DIA, as depicted in the graphs here. http://www.diabetesnet.com/diabetes-technology/insulin-pumps/pump-insulin-duration . I think you could get the IOB calculation based on the disappearance/appearance rate (derivative--subtract the current point from the point before that), although it is slightly more complex than that. Anyways, drugs in my body are making me dumb and sleepy. :P
For IOBL I guess the derivative, plus the initial insulin dose, in to a formula that has to be derived.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 22:39
So, I think for IOB using the pharmacokinetic absorption (DIA) profile, if the equation is being interpolated at t = 5 min intervals, is equal to the initial bolus dose at time t_0 minus the sum of the derivatives of the plasma insulin concentration times 5. I hope this makes sense. I really feel weird right now due to my medicine. I am comfortable though, and hopefully this medicine will keep me from ending up in the hospital again.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 22:45
Obviously the sum of the derivatives would be a completely positive value, regardless of a sign change. Actually, I am going to test it out for fun!
Scott Leibrand
@scottleibrand
Feb 21 2015 22:47
yeah, insulin activity is the derivative of IOB
when IOB drops from 1U to 0.9U, that means 0.1U of insulin has taken effect, so your insulin activity is 0.1U over whatever time period that decay took.
diabeticgonewild
@diabeticgonewild
Feb 21 2015 22:49
Yeah, that makes sense, because I was looking at a graph and it was in microunits/mL which is a concentration level. IOB would be in Units so that would be the integral of DIA (which is the derivative of IOB)
it's also in milliunits/liter in another referenced document
Scott Leibrand
@scottleibrand
Feb 21 2015 22:51
exactly
diabeticgonewild
@diabeticgonewild
Feb 21 2015 22:51
simple calculus concepts! See I'm dumb/slow today!
Scott Leibrand
@scottleibrand
Feb 21 2015 22:51
oh, not quite
integral of insulin activity, not DIA
DIA is the 4h or whatever
insulin activity units are U insulin per unit time
diabeticgonewild
@diabeticgonewild
Feb 21 2015 22:52
yeah insulin activity is absorption and DIA is an approximation right?
Scott Leibrand
@scottleibrand
Feb 21 2015 22:52
DIA units are just time
DIA is the time over which insulin activity is non-trivial
diabeticgonewild
@diabeticgonewild
Feb 21 2015 22:52
got it. I don't really care for semantics from like diabetes books and stuff, but I should. I am more in to Hovorka's models.
DIA is a "test" for insulin activity, basically
Scott Leibrand
@scottleibrand
Feb 21 2015 22:53
it's kind of a useless measure if you think about what it really means
but people use DIA as a shorthand for which insulin activity / IOB curve to use
diabeticgonewild
@diabeticgonewild
Feb 21 2015 22:53
it is, cause your pump "does it all"
Scott Leibrand
@scottleibrand
Feb 21 2015 22:53
so choosing DIA=3 vs DIA=4 gives you a curve that peaks earlier or later
diabeticgonewild
@diabeticgonewild
Feb 21 2015 22:53
correct
diabeticgonewild
@diabeticgonewild
Feb 21 2015 23:09
Actually, based off of those graphs, you would have to determine the average plasma volume for the individuals based off of the average weight provided in the document of the subjects they studied, and multiply all of the points for Insulin Absorption by it in order to get units. And then then you would have insulin concentration in the plasma in units. So, in order to get IOB it would be equal to initial bolus - the sum of the derivative points of DIA*plasma volume (graph) times time [time interval of derivative points].
I really apologize for being this dumb. Like my doctor prescribed me some medicine that's making me trip and it's making my ears ring. It's called Lomotil. If that doesn't work, which it hasn't been working ideally, my doctor is talking about prescribing me tincture of opium. Like if you Google it, it's a "no comment" issue, but it just proves that my medical problems are out of control.
I really want this medicine to work though. I don't want the other option.
Insulin Absorption*plasma volume I mean, sorry I need to get off. Drugs are bad.
Scott Leibrand
@scottleibrand
Feb 21 2015 23:12
heh. FWIW, you on drugs is about like most normal people not on drugs. :)
diabeticgonewild
@diabeticgonewild
Feb 21 2015 23:13
haha
diabeticgonewild
@diabeticgonewild
Feb 21 2015 23:53
I think for simplicity's sake the pharmacokinetics from the package inserts can be modeled for IOB this way, since plasma volume (a static constant)*insulin action graph, is virtually the same shape:
  1. Find the peak of the insulin activity graph, and set the insulin concentration value equal to 50%
  2. "Split" the graph at the peak, into two percentage brackets, 1st half 100-51% [IOB remaining], 50-0% [IOB remaining]
  3. For the left hand side of the graph, assign the minimum (initial value) as 100%. The peak will be ~51%.
  4. To find percentages for the points in between the minimum and the peak: ((peak insulin measurement in plasma - value at point)/(peak insulin measurement in plasma))/2 + 50%
  5. Do this until enough points are generated for the left hand side of the graph, with respective percentages.
  6. For the right hand side of the graph, assign a minimum (initial value)--furthest right side point) as 0% IOB. The peak will be ~50% IOB.
  7. To find percentages for the points in between the minimum and the peak: ((peak insulin measurement in plasma - value at point)/(peak insulin measurement in plasma))/2
  8. Use best fit in Excel to fit the data