We rebuilt the ranking around four health measures that every dish can be judged by fairly: processing, nutrient density, anti-inflammatory character, and glycemic load. Together, they turn a beautiful tradition into a score that is clear, comparable, and open to review.
How processed are the ingredients? NOVA is a peer-reviewed classification from unprocessed (Group 1) to ultra-processed (Group 4). Whole foods win; industrial reformulations lose hard.
Good nutrients per calorie (fiber, protein) minus penalties for sodium and saturated-fat proxy. Inspired by Drewnowski's Nutrient Rich Foods index.
Net inflammation score across ingredients — omega-3s, polyphenols, turmeric, ginger add; processed meat, refined carbs, industrial oils subtract. Derived from Shivappa's Dietary Inflammatory Index framework.
How the dish spikes blood sugar. Low-GI ingredients (legumes, whole grains, vegetables) raise the score; pure refined carbs drop it. Per Foster-Powell & Brand-Miller's international GI tables.
The NOVA classification sorts food into four groups based on the degree of industrial transformation. It's the single most predictive metric in modern nutritional epidemiology — large cohort studies (NutriNet-Santé, UK Biobank) consistently show ultra-processed food consumption as a dominant driver of cardiovascular disease, metabolic syndrome, and all-cause mortality.
| Group 1 | Unprocessed or minimally processed foods | 100 pts |
| Group 2 | Processed culinary ingredients (oils, salt, sugar, butter, honey) | 80 pts |
| Group 3 | Processed foods (cheeses, breads, cured meats, fermented vegetables) | 55 pts |
| Group 4 | Ultra-processed (industrial sausages, sweetened drinks, instant foods, reformulated products) | 20 pts |
A dish's NOVA score is the weighted average of its ingredients' group scores. Any ingredient in Group 4 applies a 20% penalty to the dish total — one industrial component shouldn't be hidden by five whole ones.
Source: Monteiro CA, Cannon G, Moubarac JC, Levy RB, Louzada MLC, Jaime PC. "The UN Decade of Nutrition, the NOVA food classification and the trouble with ultra-processing." Public Health Nutrition. 2018;21(1):5–17.
The positive signal. What does this dish deliver per calorie you spend? We use a simplified version of Drewnowski's Nutrient Rich Foods index (NRF9.3), constrained to the six nutrition values we track per dish.
Dishes rich in fiber and lean protein rise; dishes heavy in sodium or saturated fat fall. Calibrated so a lentil-forward stew lands around 70–85 and a deep-fried refined-flour snack lands around 20–35.
Framework: Drewnowski A. "The Nutrient Rich Foods Index helps to identify healthy, affordable foods." American Journal of Clinical Nutrition. 2010;91(4):1095S–1101S.
Chronic low-grade inflammation is a mechanistic pathway shared by cardiovascular disease, type-2 diabetes, many cancers, and neurodegenerative disease. The Dietary Inflammatory Index framework assigns each food an inflammation-effect score; we apply the same principle at the ingredient level.
| +3 | Strongly anti-inflammatory | Fatty fish, olive oil, turmeric, ginger, garlic, green tea, leafy greens, berries |
| +2 | Moderately anti-inflammatory | Nuts, most vegetables, whole grains, legumes, herbs |
| +1 | Mildly anti-inflammatory | Most fresh fish, alliums, cruciferous vegetables |
| 0 | Neutral | Eggs, plain dairy, chicken |
| −1 | Mildly pro-inflammatory | Lean red meat, white rice, refined flour |
| −2 | Moderately pro-inflammatory | Cured meats, industrial seed oils, refined sugar |
| −3 | Strongly pro-inflammatory | Processed/packaged meats, high-fructose corn syrup, trans fats |
A dish's score is the mean inflammation index of its ingredients, mapped to 0–100. Editorial approximations — not verbatim published DII values, which are parameter-level (45 food components), not ingredient-level.
Framework: Shivappa N, Steck SE, Hurley TG, Hussey JR, Hebert JR. "Designing and developing a literature-derived, population-based dietary inflammatory index." Public Health Nutrition. 2014;17(8):1689–1696.
Glycemic response — how fast a meal pushes blood glucose up — predicts metabolic health independent of total calories. We tag each ingredient with a low/medium/high flag using published GI reference tables.
| Low GI (< 55) | Legumes, most vegetables, dairy, nuts, seeds, most fruits, quinoa, pearl barley | 100 pts |
| Medium GI (55–69) | Basmati rice, whole-wheat bread, oats, couscous, sweet potato | 65 pts |
| High GI (≥ 70) | White bread, white rice, potato, refined flour, high-fructose corn syrup | 30 pts |
Score = weighted mean across carb-bearing ingredients. If more than half the carb ingredients are high-GI, the dish is penalised an additional 25%. Protein- and fat-only dishes (no significant carbs) default to 88 — they don't spike glucose, but they aren't blood-sugar winners either.
Reference: Foster-Powell K, Holt SH, Brand-Miller JC. "International table of glycemic index and glycemic load values: 2002." American Journal of Clinical Nutrition. 2002;76(1):5–56. Subsequent updates via Harvard Health Publishing GI/GL tables.
Neither of these affects the rank. They're filters: sort the top 2,024 by health, then narrow to dishes you can actually make on a weeknight or afford on a grocery budget.
Pure ingredient count. A rough proxy for "how many trips to the spice shop".
Abstract buckets based on the rarest ingredient in the dish. No dollar amounts — saffron is "Luxurious" in every country. The most expensive ingredient sets the tier.
python tools/build_data.py and you get the same result.2,024 dishes / 199 countries. Methodology unchanged from v4.0; corpus audited (100 same-dish-different-slug duplicates removed; country-name canonicalization fixed Czechia/Côte d'Ivoire/Palestine merges; Hawaii reclassified to United States). Recipe-link feature removed for editorial focus.
Corpus grew 897 → 1,944 dishes through three authoring volumes — Vol IV (+300, gap-fill for thin countries), Vol V (+300, ultra-regional + microstates), Vol VI (+500, real cooked dishes only — drinks/sauces/condiments stripped).
Health is the only score. Health = 35% NOVA + 25% Nutrient Density + 20% Anti-inflammatory + 20% Glycemic Load. Popularity and Heritage removed. Simplicity and Affordability added as filter badges.
Overall = 80% Health + 10% Popularity + 10% Heritage. Health score was
editorial judgment, not a reproducible formula. Deprecated because heritage
ranked across cultures wasn't defensible and popularity had no data source.
Snapshot kept in /archive/v3.0/.
GastroRank is an editorial and educational project that applies public nutrition frameworks to traditional dishes. It is not medical advice, nutritional counselling, diagnosis, treatment, or a substitute for care from a qualified health professional. Scores are comparative estimates based on typical ingredients and serving profiles; individual needs, allergies, medical conditions, medications, preparation methods, and portion sizes can materially change what is appropriate for you.
Do not use these rankings to make clinical, regulatory, commercial, or personal health decisions without consulting an appropriate professional.
Monteiro CA, Cannon G, Moubarac JC, Levy RB, Louzada MLC, Jaime PC. The UN Decade of Nutrition, the NOVA food classification and the trouble with ultra-processing. Public Health Nutrition. 2018;21(1):5–17.
Shivappa N, Steck SE, Hurley TG, Hussey JR, Hebert JR. Designing and developing a literature-derived, population-based dietary inflammatory index. Public Health Nutrition. 2014;17(8):1689–1696.
Foster-Powell K, Holt SH, Brand-Miller JC. International table of glycemic index and glycemic load values: 2002. American Journal of Clinical Nutrition. 2002;76(1):5–56.
Drewnowski A. The Nutrient Rich Foods Index helps to identify healthy, affordable foods. American Journal of Clinical Nutrition. 2010;91(4):1095S–1101S.