USDA FoodData Central
The U.S. Department of Agriculture's FoodData Central is the gold standard for nutrition data in the United States. It includes:
- Standard Reference Legacy (SR Legacy): ~7,800 foods analyzed by USDA laboratories, with detailed macro and micronutrient profiles.
- Foundation Foods: ~200 frequently consumed foods with the most comprehensive nutrient analyses available.
- Survey (FNDDS): ~7,500 foods from national dietary surveys, including many prepared and mixed dishes.
- Branded Foods: over 1.6 million branded items with label-derived nutrition.
We use the USDA's official API and update our cached nutrient values regularly. USDA data is in the public domain and free to use.
Open Food Facts
For international branded products, we supplement USDA data with Open Food Facts — a community-built, open-source database of food products from around the world. Open Food Facts is licensed under the Open Database License (ODbL), which we comply with by attributing the source and contributing improvements back.
How we handle missing or conflicting data
- Missing nutrients: when a food's profile doesn't include a specific nutrient (say, vitamin K isn't measured), we display "—" rather than guessing or zeroing.
- Conflicting values: if USDA and Open Food Facts disagree on a branded item, we prefer USDA's measurement. Where USDA doesn't cover a specific brand, we use Open Food Facts and clearly mark the source.
- Rounding and units: we display values rounded to a sensible number of digits. Underlying calculations use full precision.
What about portion estimation?
When you describe a meal in plain language, Lygo's AI estimates portion sizes based on typical human descriptions ("a bowl of," "a slice of," "a handful"). These are informed estimates, not measurements. If you want precision, weigh your food or describe weights directly ("60g of oatmeal").
Reporting an error
Spotted a nutrient value that looks wrong? We'd love to fix it. Email hello@lygoapp.com with what you saw and the food you logged. We trace it back to the underlying source and either correct our data or report the issue upstream.