World Table data methodology

What data is used

Every figure comes only from meals whose owners explicitly consented to share them. Private meals, withdrawn posts and posts under review never enter any aggregate.

Per-food nutrition uses only that food item’s own values. The calories of a full meal are never copied onto its rice or side dishes.

Why medians instead of averages

Portion sizes vary widely between people. Averages are easily dragged by one or two unusually large or small meals, so we report the median (P50) and the 25–75% range.

A sample is never dropped just for being large. A genuinely big meal is still real data.

What gets excluded

Only mechanically detectable errors: negative or non-numeric values, non-positive total weight, unsupported units, nutrition totals that do not add up, and stale values from before a user edited the meal.

Per-100g figures are computed only from samples with reliable weights, and are shown only once at least 5 such samples exist.

How many meals before numbers appear

Nutrition figures are published once 5 or more valid shared meals exist. With 1–4 meals we show the meal cards and description only.

If a food drops to zero shared meals, its page is removed from search indexing.

What does the colour of the estimated post-meal peak mean

The number shown is an AI estimate of the post-meal peak, from a 100 mg/dL fasting baseline. Its position within a 70–250 mg/dL scale is shown as an interpolated colour, with no status label such as “safe” or “caution”. Cutting the range into a few bands pushes different meals into the same band and erases the difference, and a label reads like a clinical verdict.

The points where the colour changes follow post-meal targets from published guidelines. The International Diabetes Federation (IDF) gives a 1–2 hour post-meal target of 140 mg/dL, and 160 mg/dL or below for people with diabetes. The American Diabetes Association (ADA) gives below 180 mg/dL 1–2 hours after a meal for adults, and the U.S. Centers for Disease Control and Prevention (CDC) cites below 140 mg/dL two hours after eating as a general reference.

These figures are population-level targets, not a test applied to an individual. Post-meal responses to the very same food have been reported to differ markedly between people (Zeevi et al., Cell, 2015). The colour is therefore only a marker of position on the scale, and should not be read as a verdict that a particular value is good or bad.

AI estimates vs measured glucose

The glucose impact distribution is an AI estimate based on food composition, not a measurement. Real responses vary greatly with individual condition, activity and medication.

For meals that do have a measured record, the public web shows only that a record exists and how many such cases there are. Individual values, deltas and timings are never shown publicly, and no per-food measured averages or ranges are produced.

Disclaimer

Everything here is general information, not medical advice. Always consult a healthcare professional for diagnosis, treatment or medication decisions.