🥗 One Good Choice

The Food Compass Score (FCS) evaluates foods across 9 distinct domains, each measuring different aspects of nutritional quality. All measurements are normalized per 100 kilocalories for fair comparison.

How to Read the Chart

Each domain is displayed as a diverging bar chart:

• Green bars extending right = Positive scores (beneficial)
• Orange bars extending left = Negative scores (harmful)
• Center line = Neutral point (score of 0)
• Bar length = Magnitude of impact

The 9 Domains

Click the i button next to any domain name to learn what it measures.

• Nutrient Ratios - Balance of fats, fiber, and minerals
• Vitamins - Essential vitamin content
• Minerals - Essential mineral content
• Food Ingredients - Whole vs processed ingredients
• Additives - Added sugars and preservatives
• Processing - Level of food processing
• Specific Lipids - Beneficial and harmful fats
• Fiber & Protein - Key macronutrients
• Phytochemicals - Beneficial plant compounds

The overall FCS score (1-100) is calculated by combining all domain scores with the calorie density of the food.

This Food Compass Score is calculated using 41 specialized neural network models that predict individual nutrients and food components from the food's name.

🧠 What Gets Predicted?

• Macronutrients: Calories, protein, carbs, fiber, fats
• Vitamins: A, C, E, B6, B12, folate, thiamin, riboflavin, niacin, vitamin K
• Minerals: Sodium, potassium, calcium, iron, magnesium, phosphorus, zinc, copper, selenium
• Lipids: Saturated fat, unsaturated fat, cholesterol, omega-3s EPA/DHA/ALA
• Food Groups: Fruits, vegetables, whole grains, refined grains, nuts, seafood, red meat, beans
• Processing: NOVA classification, added sugar
• Phytochemicals: Flavonoids, Carotenoids

⚠️ What Uses Defaults?

A small number of nutrients still use conservative default values (typically 0) due to limited training data or model compatibility issues:

• Vitamin D (limited food sources, harder to predict)
• Choline (model file mislabeled)
• Some food group categories (dried fruits/vegetables, plant oils, MCT oils, yogurt)
• Some processing indicators (cured meats)
• Fermentation, frying, nitrite indicators (insufficient training data)

Why This Matters: Foods rich in the missing nutrients may score slightly lower than they should. For example, fatty fish (vitamin D), olive oil (plant oils), or fermented foods (fermentation indicator) might be undervalued by 5-15 points. We're continuously training new models to improve coverage.

Transparency First: We believe in showing you exactly what's predicted vs. estimated. No black boxes, no hidden defaults. What you see is what the AI actually knows.