I’m wondering if there are any easy-to-use tools that allow you to visualize how much insulin is active in the body for any given 5-minute window. Not the amount of insulin still on board – which could reflect insulin available over the next two to three hours at least – but just an estimate of how much insulin is actually acting to lower blood sugar in the moment.
My knowledge of calculus is very limited (and old) but I remember that the calculus integral function continuously adds up the many slices of the “area under the curve,” in this case insulin action over time. I think that calculus could also calculate one five minute-slice of that at a particular time and yield the insulin quantity answer you are looking for.
The people who make up the brains behind the “we are not waiting” movement could probably supply this answer, if they haven’t already.
Any answer that math could provide would just be an ideal approximation since actual insulin action will vary from person to person and even in the same person from meal to meal. A static formula would be a “best guess.”
That’s an interesting question.
YEah, I think that I could probably calculate it if need be. What I’m looking for is some sort of tool to help me visualize it as a kind of line/bar graph. I think that will help explain some of the befuddling trends I keep seeing in my son’s data – for instance, he tends to dip low at a few points in the day. Now, it could be that his necessary basal rate just drops precipitously at these times, but I suspect it is more likely about residual insulin, slightly mismatched basal rates earlier in the day, etc. I could try to figure this all out with some kind of painstaking variable isolation – basal testing, varying how many carbs he eats per meal, etc. – but I’ve realized that it’s probably easier if I can just see the actual amount of insulin (both total basal and bolus), and I’m guessing the answer will just reveal itself.
I guess I have to hassle my husband to develop this tool.
If you and your husband do this, please update here. Many people don’t get into that level of the nitty-gritty but instead simply observe the symptoms and factor them into what they call the “art” of insulin dosing. Operating from this fuzzier “intuitive” perspective can still be effective.
1 Like
Yes, this fuzzy form of insulin surfing is often what I do and when I am with my son it works well! The problem is that I basically am hoping to wean down to just popping in to daycare just once a day – for lunch insulin dosing – and given that I am simply unable to just let him be super high, I need to come up with a system that works more on autopilot than this method.
Tia, unfortunately this is quite a bit more complicated than you think for what you’re trying to accomplish.
At any given moment, BG concentration in the blood is a function of two factors: Glucose ingress, and glucose removal (mediated by insulin).
So simply know insulin levels in the blood isn’t enough to figure out why BG levels are what they are (r why they’re changing, etc.). You also have to have a good idea of what sugar entering the blood – from digestion, from glycogenolysis in the liver, fatty acid metabolism, etc.
1 Like
Dave26 – yes, for sure other factors are at play! But I already have a rough sense of at least some other factors like how much my son is eating at given times and I *think I know the carb-absorption rate – especially for a low-fat meal like oatmeal. While I don’t know his liver output, I do have a rough sense how much his liver is releasing when, at the same time, he doesn’t eat (because of weekends).
So maybe a visualization like this may not reveal all the secrets to my son’s BG trends. But in this particular case, I suspect insulin stacking is being hidden because the basal rate is slightly too high for some, but not all, of the morning. If basal rates are not completely right at all times, then the IOB calculation will not reveal this type of stacking.
I definitely think it will be illuminating. I also think it’s actually a lot less complicated than spending two or three days basal testing, two or three days slowly ramping up/down carbFs, or shifting basal increases right or left by 30 minutes.
Should be pretty easy to calculate.I am not sure, however, how much more you would be able to learn from such insulin action plot beyond what you already see from the bg graph, IOB, and openaps prediction curve.
I’ve tried using all the NS and openAPS data to figure out where these drops come from and it’s not apparent with a glance. Perhaps I’m just being dense though!
I don’t think you can do it as the derivative of IOB, because openAPS is using basal as a floor, so IOB doesn’t account for your set basal rate. For us this is a huge issue as our son’s basal rates are changing so often that we, for instance, don’t have smooth transitions between highs and low rates (we depend on openAPS to pick up the slack) and we know there are times when it’s off because it’s often compensating for food.
I think you’d need to do it by drawing directly from the reservoir value. But that, too, only reveals how much insulin is being delivered. You’d then need to use some sort of insulin action curve/function (which will be a guesstimate based on your DIA) to figure out what fraction is active in any 5-minute increment.
I agree - just derivative of IOB would not work. Using the reservoir values, and an insulin action curve would be the best way to go.
A delta in bg is due to a difference between two numbers, insulin action and a net glucose action (sum of all opposing effects) at the time. The difference is the problem. To compute any one of the two, you have to know or model the other. We have at least some idea about the insulin action curve and DIA, while carb absorption, or whatever liver may be doing, or whatever effects any leftover beta cells may be at any time is a total guess. I think it would be very interesting to compute insulin action as you suggested, based on reservoir values, assumed action curve and DIA, and then use that together with bg deltas to compute net glucose action curves. Would probably need to apply some smoothing to minimize noise. The net glucose action curves could then be used to potentially automate basal adjustments, or to better model carb absorption curves for particular types of meals.
I hate the IOB calculated by Omnipod I had always thought that stack-able progress bars would make more sense. Maybe even color coded so that peak effectiveness would be a different color. This would make it easier in my mind to see what the IOB is when there are multiple boluses in effect at the same time.