An independent research collaboration on the validity of dietary assessment instruments.

We study how dietary assessment instruments — food-frequency questionnaires, 24-hour recalls, weighed-food records, and the newer generation of image-based and AI-mediated tools — perform against weighed-food reference, and how their reported accuracy compares with the methodology used to produce it.

Our publications are released open-access under CC BY 4.0. We do not accept funding from, and do not enter commercial relationships with, any of the dietary assessment products evaluated in our work — see our funding statement and conflict-of-interest policy.


Recent publications

All publications →


What we study

  • Validation of image-based and AI-mediated dietary assessment
  • Agreement statistics for nutritional intake measurement
  • Vendor-reported vs. independently-replicated accuracy
  • Cuisine and population coverage in benchmark test sets
  • Reporting standards for digital health applications

How we work

The Initiative is small. We pre-register protocols, publish weighed-food protocols and code, prefer Bland-Altman and equivalence testing over null-hypothesis significance for agreement claims, and report limits of agreement as our primary outcome for accuracy work. Where we evaluate commercial products, we evaluate the public-facing surface as a black box; we do not enter commercial NDAs.