Answers are grounded in this paper and cite their sources.
Background: Continuous glucose monitoring is usually summarized by mean glucose, variability, and time in range, which do not separate meal responses from between-meal drift. Methods: We analyzed 125 recordings from 112 adults with diabetes, holding 3470 meals with recorded times and weighed contents. Known meal times let baseline and meal response be estimated together from the whole trace, without a washout. Results: The fitted displacement was 32.9 mg/dL, peaking 105 min after the meal, against 45.4 mg/dL for the 120 min maximum rise and 27 to 35 for the excess over shuffled meal times. The response cleared 4.8 times faster than the baseline returned, rejecting the instantaneous jump in 100 of 117 recordings. Its assumed independent error failed in all 117 recordings. Allowing 90 min of memory returns a displacement 5 times smaller. On simulated recordings with that memory the first-order estimator recovers a true 45.0 mg/dL and a true 6.3 alike, while the longer-memory one turns negative, so the smaller figure measures the estimator. It does correct the half-time, 1.9 h against 6.7 h. Within patients the displacement rose with staple food mass, not meal weight. Conclusions: The parameters did not increase glycated hemoglobin prediction (change in R2 0.020, −0.035 to 0.084).
No takes yet. Share an insight, caveat, or question.
Naskinova et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: