trying to sanity check something before i waste another week of logs on the wrong frame. i’ve been logging fasting glucose and a 1-10 hunger score by day-of-week since around 7.5mg, partly to see if i could see the drug ‘wearing off’ before the next shot. and the pattern i thought i was going to see (smooth ramp down day 4-5, noticeable drop day 6-7) isn’t really there. it’s more like… flat-ish all week with noise i can’t cleanly attribute to anything. which made me go back and actually read on the PK instead of assuming. half-life is around 5 days, so at steady state day 6-7 isn’t a clean trough, it’s something like 60-65% of peak. that’s not ‘the drug wearing off,’ that’s still pretty solid receptor occupancy. so my actual question: for people who track by day-of-week, are you finding day 6-7 readings meaningfully different from day 2-3, or is it mostly noise inside a fairly flat curve? trying to figure out if the day-of-week framing is even the right axis or if i should be looking at something else (dose week number, vial age, sleep the night before). the journal note field tied to each dose entry is what’s let me even ask this honestly, otherwise i’d be backfilling a story onto the numbers. fwiw i’m pre-diabetic not T2D so ymmv.
the PK read is right but you’re testing it on the wrong dependent variables. fasting glucose at steady state is going to be flat-ish bc hepatic glucose output is suppressed across the whole dosing interval, and a 1-10 hunger score is too coarse to catch a 30-35% trough-to-peak swing in receptor occupancy. if you want to see the curve, postprandial AUC on a standardized meal is way more sensitive, and MAGE from CGM if you have one would show it cleaner than either of those. day-of-week as an axis isn’t wrong, it’s just that fasting glucose smooths over what you’re trying to measure. vial age is the variable I’d add before sleep tbh, drift there is easier to catch than people think.
the axis i’d add that nobody’s named yet: are you actually at steady state on 7.5 yet, or still ramping into it? with a ~5 day half-life it takes roughly 5 half-lives to plateau, so ~25 days, call it 3.5 weeks of unchanged dose before the week-over-week baseline stops climbing on its own. if you bumped to 7.5 anywhere inside that window, your “day 6 vs day 2” comparison is riding on top of an accumulation ramp, and the ramp is a bigger signal than the trough-to-peak wobble you’re hunting for. so dose-week-number isn’t just another candidate axis, it might be the confound silently flattening the day-of-week one, because the slow rise across weeks partially cancels the within-week decay you expected to see. separate thing, and i’ll flag this as my own extrapolation since the public stability data doesn’t really license a hard number: within-vial position. dose 4 out of a multidose vial isn’t dose 1 chemistry even on an identical fridge log, headspace ratio climbs as it draws down. that’s a different variable than t.rousseau’s vial-age point, which is calendar drift, not draw-order drift. probably third-order behind steady state and meal-response sensitivity, but if you’re already logging vial age you’re one field away from logging draw number too. for what it’s worth the flat-ish fasting glucose tracks with what i’d expect at 7.5 pre-diabetic, mine stopped showing day-of-week structure once i was past the titration ramp, the variance left over read more like sleep and prior-day carb load than anything dose-phase. tagging each entry with which absolute dose-week it was, plus the check-in reminder so i wasn’t reconstructing the date later, is what let me separate the ramp weeks from the steady ones after the fact. i’d lock down “am i at steady state” before spending another week on the day-of-week axis, otherwise you’re measuring two curves stacked on each other and calling the sum noise. ymmv since n=1.
The steady-state point is the right thing to lock down first, no argument there, and tagging absolute dose-week so you can pull the ramp weeks out afterward is exactly the move. Where I’d ease off is the within-vial draw-order idea doing real work here. The case for it is fair on paper, headspace ratio does climb as you draw a multidose vial down. But that hypothesis makes a specific prediction: drift should be monotonic across draws 1 through 4 on an otherwise identical fridge log. When I chased an 8 mg/dL CGM drift over two weeks, it traced back to my own reconstitution inconsistency, not vial age and not draw order, and the tell was that it wasn’t proportional to anything. The non-linearity is what ruled out simple first-order decay, and it’d rule out a clean headspace ramp too. So before you add draw-number as a field, I’d ask what you can actually validate it against. Without a potency assay on each draw, draw-number is loggable but not checkable, and a field that’s there because it’s easy to fill rather than because it carries signal will quietly over-explain your noise. I do keep the per-vial fields next to the glucose log for exactly that tie-back reason, and the overlay where you can stack two trend lines is what makes a drift visible before I’d otherwise spot it, but I treat draw-order as a flag to revisit, not a variable I trust yet. And since you’re pre-diabetic, fasting glucose is the least sensitive axis you’ve got for any of this. A postprandial reading or the hunger-score curve would move first if anything’s moving at all.