Analyses

The evaluation covers various analyses of a food selection. All analyses are based on the nutrient data and classifications of the generic single foods. This is possible because both generic composite foods and branded foods in our database ultimately resolve to recipes composed of generic single foods.

Each analysis requires a list of foods with consumption amounts (e.g., a day log or a meal) in the request to the respective endpoint. For foods with a processing state, we take into account whether the consumption amount refers to the original form or the processed state. If the food was weighed before preparation, weight changes and nutrient losses are applied.

Record time of weighing

In dietary surveys, the boolean flag weighted_before_process (weighed before preparation) should be asked when entering the food item.

Nutrition Summary

POST /evaluate/nutrition-summary

The nutrition summary returns the sum of all nutrients consumed across the selected foods. This corresponds to a complete nutrient calculation of the diet. In addition to calories and macronutrients, you can also analyze vitamins, minerals, and other constituents.

Don't overinterpret the nutrition summary

The accuracy of micronutrient values strongly depends on data availability in BLS 4.0. Analyses such as the Healthy Eating Index and energy density are preferable.

Macronutrient Distribution

POST /evaluate/macro-distribution

Macronutrient distribution refers to how total energy is split across macronutrients. Specifically, it calculates what share of calories comes from carbohydrates, fats, proteins, and alcohol. The result is reported as a percentage of total energy and gives a quick overview of the diet’s macronutrient balance.

Don't overinterpret macronutrient distribution

Even with identical macronutrient distribution, overall dietary quality can differ substantially. Analyses such as the Healthy Eating Index and energy density are preferable.

Healthy Eating Index

POST /evaluate/healthy-eating-index

The metric Healthy Eating Index (mHEI) is a published variant of the U.S. Healthy Eating Index (HEI). It assesses diet quality based on the Dietary Guidelines for Americans 2015. The HEI originated in the U.S. and uses U.S. units and reference portions. The mHEI adopts the 13 scoring components of HEI-2015 unchanged but converts all quantities to the metric system. The core HEI framework has not changed since 2015 and remains current.

Component overview

Component	Description	Points
total_fruits	Fruits, total	5
whole_fruits	Fruits excl. fruit juice	5
total_vegetables	Vegetables, total	5
greens_and_beans	Dark-green vegetables & legumes	5
whole_grains	Whole grains	10
dairy	Dairy & plant-based alternatives	10
total_protein_foods	Protein foods, total	5
seafood_and_plant_proteins	Seafood & plant proteins	5
fatty_acids	Fatty acid ratio (PUFA+MUFA)/SFA	10
refined_grains	Refined grains	10
sodium	Sodium	10
added_sugars	Added sugars	10
saturated_fats	Saturated fats	10
total_score	Total score	100

A high mHEI score indicates adequate intake of fruits, vegetables, whole grains, dairy, protein foods, and healthy fats (adequacy components) and not too much refined grains, sodium, added sugars, and saturated fats (moderation components).

Instead of focusing on individual nutrient amounts, the index emphasizes food groups and dietary patterns relative to energy intake. This highlights the importance of a varied selection of primarily nutrient-dense, low-energy foods. Absolute energy is not decisive for the calculation; rather, it’s the quality of the selection relative to its energy content. The mHEI is therefore applicable at different scales—from day- or week-level logs to entire recipe databases. It is less suitable for single meals, as it targets overall dietary patterns.

Use component scores

While the total mHEI score provides a quick overview of diet quality, the individual component scores should be prioritized in counseling contexts. They help derive concrete opportunities for improvement.

Energy Density

POST /evaluate/energy-density

Prototype stage

The energy density analysis is still experimental. There is currently no universally accepted scientific method to automatically compute energy density for an entire diet from database values without losing context.

Development outlook

We plan to evaluate and, if needed, further develop the calculation logic.

Energy density (ED) describes how much energy a food provides per unit mass (kcal/g). Applied to the total diet, average ED indicates how energy-dense the selection is on average. Beverages are excluded because they often have very low ED and are less satiating than solid foods. Even highly sugary drinks like sodas would typically lower the ED of an entire diet. Their calories are therefore considered separately (“beverage calories”).

Challenges with dietary logs

Edge cases arise when items are logged separately even though they were consumed together. For example, if a coffee with sugar is logged via generic single foods as “Coffee (beverage)” and “Sugar.” In that case, sugar would count as a solid toward ED, though in context—similar to “Coffee (beverage) with sugar” from generic composite foods—it should not. This inaccuracy can only be avoided by additional prompts during logging and must be weighed against the extra effort.

The split between “counts as food” and “counts as beverage” is not trivial in practice. Liquids or semi-liquids such as smoothies, soups, or milkshakes can be considered food or beverage depending on context. We therefore classify generic single foods in the database as is_solid = True/False. For generic composite foods and branded foods, we derive the is_solid flag from the included generic single foods. If more than 25% of the ingredients (by weight) are solids, the entire item is classified as is_solid = True. Hydration factors for branded foods are taken into account.

Account for processing state

For reliable computation, processing states and weighted_before_process should always be provided. Otherwise, the analysis can be heavily distorted (e.g., by pasta logged in its raw form).