How Pursira uses data and research

Nutrition data

When you search for a food or look up nutrition information while logging a meal, Pursira can use USDA FoodData Central to help find matching foods and their nutrition details. Not every food record in Pursira necessarily comes from USDA -- other data sources or locally stored entries can exist alongside it -- but USDA FoodData Central is the external nutrition database Pursira's food search integrates with.

USDA FoodData Central groups records into different data types, including Foundation foods, Survey (FNDDS), SR Legacy, and Branded products, and Pursira's food search considers all of these. Because of this, the nutrition values you see depend on which record you select: a generic ingredient entry and a branded product with a similar name can report different values, since manufacturers submit their own nutrition facts for branded items. When a food is logged at a different serving size, Pursira scales the reported values but does not alter USDA's underlying nutrient data.

Activity calorie estimates

When you log an activity -- such as walking, running, cycling, hiking, swimming, or rowing -- without entering a calorie value yourself, Pursira can estimate the calories burned using MET (metabolic equivalent) values. The current implementation references the 2024 Compendium of Physical Activities for these MET values. These are estimates, not direct measurements: actual energy expenditure varies between individuals in ways a MET-based estimate can't fully capture.

A MET value describes how many times more energy an activity takes compared with resting; multiplying it by your body weight and duration produces an estimated calorie total. This is an estimate, not a direct measurement -- it doesn't account for fitness level, terrain, effort, or body composition. If you enter your own calorie value, Pursira uses that instead of estimating one.

Research-informed AI Coach

Pursira maintains a curated corpus of published research that the AI Coach can draw on when answering research-related questions. When an answer is grounded in this research, the AI Coach can show you the citations it used -- including each paper's title, authors, journal and publication year, a DOI or source link, and license information. The goal is to let you inspect the underlying source yourself rather than asking you to trust an unsupported claim.

When you ask a research-related question, Pursira searches its research corpus for entries that match your question, and only that retrieved evidence may support research claims in the answer -- earlier conversation messages aren't treated as evidence on their own. The corpus is a small collection, not a comprehensive literature review, so citations only appear when matching research was actually retrieved; many questions won't have one.

Try the AI Coach after signing in to see research citations in practice.

What Pursira develops itself

Not everything in Pursira comes from an external standard. Some parts are Pursira's own designed product logic:

• Exercise catalog wording. Instructions, common mistakes, and precautions for each exercise are written in Pursira's own original wording, not copied from an external source.
• Recovery and readiness scoring. How Pursira turns your recovery check-in into a readiness score is a heuristic Pursira designed -- it is not a clinical or medically validated assessment. It converts self-reported inputs, such as sleep and how recovered you feel, into a score using rules Pursira wrote, and it does not diagnose fatigue, illness, or injury.
• Workout recommendations and exercise ranking. Which exercises get prioritized for a given muscle group reflects Pursira's own product decisions, informed by general training principles rather than a single peer-reviewed standard. This ordering was developed by reviewing publicly available exercise-ranking content and general resistance-training principles, then synthesizing Pursira's own prioritization -- it is not copied from a single source and isn't itself a scientific study. The Recommended option can also factor in your saved profile (goal, fitness level, equipment), recent training history, and, for a workout planned today, your most recent recovery check-in, when available.

None of this is peer-reviewed or a clinical standard -- it's how Pursira chose to build the product.

What these sources do not mean

• Referencing external data and research does not make Pursira a medical service.
• Activity and calorie values shown in Pursira are estimates, not precise measurements.
• Recovery and readiness scores reflect a self-reported check-in, not a medical assessment.
• Fitness recommendations are meant to support your training, not guarantee a specific outcome.
• Research-based AI Coach answers reflect only what is in Pursira's current research corpus, not the entire body of scientific literature.
• For medical concerns, talk to a qualified healthcare professional.