Shooter Guide to Lantus
I've made a few references in this series to "How much Lantus to shoot?" Here's how I do it. If you're squeamish about syringes and injections, you might want to stop here. For the rest of us, let's start with a graph ...
A few definitions: The data is for the prior 21 days, and the chart is updated daily. The y-axis is "ON_drop", overnight drop, in my BG measure the morning after the shot. The x-axis is "Shot Size", or how many units of Lantus I drew into the syringe. Excel does the simple linear regression line, and provides the slope, intercept, and R-squared values. The blue bars in upper left are used to position the red lines on the chart. The length of the blue bars is the standard deviation of the last 21 days of overnight drop data.
The prior 21 days - 3 weeks - is the data of interest because I assume my insulin sensitivity will change over time, and I want to capture that and include it in the model.
Every night, I need to decide how much Lantus to draw. So, I do my final BG measure of the day. My target for the morning BG measure is 100, so final BG - 100 = targeted ON_drop. Then using the chart, and the regression line, I get an approximate value for shot size. I apply windage to that approximate value, because I want to put as little Lantus in me as possible. Less is better, and more risks hypoglycemia. I usually round down to zeros (30, 40, ...) or fives (35, 45, ...) because it's doable on the syringe barrel, and it makes the chart easier to read. I can get to ones if needed.
A few final notes: Some of the numbers on the chart are negative, showing a BG increase overnight. That is an interesting subset of the data, because those numbers are almost "the same". For some values of Shot Size, the range of responses is pretty big (30: 90 to -18, 40: 113 to -15, 55: 105 to -9). The range is about 2SD. My guess at the mechanism is the content of my stomach at the time of the "last BG". Throw in windages from time of last meal, and time of first BG measure, and dawn effect, and sleep quality.
My expectation is that continuous glucose monitoring (CGM) devices, and their siblings (ketone, trigs, et al), will eventually feed their data to an AI system that will do all this stuff for ordinary people. Some of this is already in use. See https://blog.virtahealth.com/videos-conference-science-carbohydrate-restriction-ketosis/
Finally, this is all my own work. And while I'm pleased that I could do it, I'm pretty resentful that my medical support team isn't useful in this regard. Except for the great Librarian at the Werner Medical Library.
 |
| Targeting Shot Size graph, November 3, 2018 |
A few definitions: The data is for the prior 21 days, and the chart is updated daily. The y-axis is "ON_drop", overnight drop, in my BG measure the morning after the shot. The x-axis is "Shot Size", or how many units of Lantus I drew into the syringe. Excel does the simple linear regression line, and provides the slope, intercept, and R-squared values. The blue bars in upper left are used to position the red lines on the chart. The length of the blue bars is the standard deviation of the last 21 days of overnight drop data.
The prior 21 days - 3 weeks - is the data of interest because I assume my insulin sensitivity will change over time, and I want to capture that and include it in the model.
Every night, I need to decide how much Lantus to draw. So, I do my final BG measure of the day. My target for the morning BG measure is 100, so final BG - 100 = targeted ON_drop. Then using the chart, and the regression line, I get an approximate value for shot size. I apply windage to that approximate value, because I want to put as little Lantus in me as possible. Less is better, and more risks hypoglycemia. I usually round down to zeros (30, 40, ...) or fives (35, 45, ...) because it's doable on the syringe barrel, and it makes the chart easier to read. I can get to ones if needed.
A few final notes: Some of the numbers on the chart are negative, showing a BG increase overnight. That is an interesting subset of the data, because those numbers are almost "the same". For some values of Shot Size, the range of responses is pretty big (30: 90 to -18, 40: 113 to -15, 55: 105 to -9). The range is about 2SD. My guess at the mechanism is the content of my stomach at the time of the "last BG". Throw in windages from time of last meal, and time of first BG measure, and dawn effect, and sleep quality.
My expectation is that continuous glucose monitoring (CGM) devices, and their siblings (ketone, trigs, et al), will eventually feed their data to an AI system that will do all this stuff for ordinary people. Some of this is already in use. See https://blog.virtahealth.com/videos-conference-science-carbohydrate-restriction-ketosis/
Finally, this is all my own work. And while I'm pleased that I could do it, I'm pretty resentful that my medical support team isn't useful in this regard. Except for the great Librarian at the Werner Medical Library.
Comments
Have you thought about combining this data with your other data set that shows daily variations on your BG level? You might end up with a Saturday curve and a Thursday curve...
Keep up the good work!!
My "target" for BG on wakeup is 100, each day, every day, and so I subtract 100 from my pre-bed BG. I don't see much day day variability in specific day numbers for wake up BG or average of all day BGs. I'll look at the night time ones, and see if there's anything to that. Hey, it's another blog post subject !!
Thanks for the comments.
rr