Back

Short-term postprandial glucose monitoring reveals stable traits from noisy free-living meals

Toumi, M.; Salathe, M.

2026-09-15 nutrition
10.64898/2026.09.14.26362991 medRxiv
Show abstract

Postprandial glucose responses differ greatly between meals, which complicates the identification of individual glucose response traits in free-living conditions. We analysed 50,463 free-living meals from 992 adults without diagnosed diabetes using a multivariate mixed-effects machine-learning framework and evaluated transfer to 4,524 standardized meals excluded from training. This model separated the population-shared from participant-specific patterns across four joint glucose response outcomes. Within-individual variability dominated free-living responses, yet the between-individual variation was concentrated in two dimensions. The dominant axis represented an overall tendency toward larger glucose excursions. The second axis reflected carbohydrate responsiveness with lower late glucose elevations. Dominant-axis scores were reproducible across non-overlapping meal subsets and associated with held-out standardized-meal responses, beyond age, sex, BMI, and glucose mean and variability. Dominant-axis scores estimated from 3 days of monitoring approximated those estimated from 14 days. These findings support short-term monitoring to measure glucose response individuality and identify a candidate coordinate for stratification.

Matching journals

The top 2 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.