Back

Returning Genomic Dietary Data to Participants by Translating Stool DNA into a Personal Food Profile

Jeffrey, K. R.; Rivera, C. N.; Aqeel, A.; Gedye, M. J.; Ives, N. W.; Jiang, S.; Kirtley, M. C.; Bauer, A. E.; David, L. A.

2026-07-31 nutrition
10.64898/2026.07.29.26359128 medRxiv
Show abstract

Returning individual results to research participants is increasingly expected in genomic studies. Yet, genomic dietary data have unique characteristics. New DNA sequencing techniques reconstruct diet from the residual plant and animal DNA recoverable in stool, reporting not nutrients or calories but a list of the species a person ate. Turning that list into something a participant can understand does not yet have an established framework. The Everyone EATS (Edible Atlas Through Sequencing) study returned personalized genomic dietary profiles to a non-clinical pilot cohort. Participants collected a stool sample at home, samples were processed with FoodSeq, and each participant received an interactive Diet Data Return Report that translated detected taxa into familiar food groups and benchmarked them against prior FoodSeq cohorts. No monetary compensation was offered, and the Diet Data Return Report was the sole incentive. Of 111 kits mailed, 80 were returned (72.1%) and sequenced (100% sequencing success for plant DNA; 98.7% for animal). Among report recipients with engagement data (n = 73), all accessed their report and 74% engaged interactively with its content. Among 16 feedback survey completers, 13/16 (81.2%) found the report easy to understand, 13/16 (81.2%) reported improved understanding of their dietary variety, and 5/16 (31.2%) reported a dietary behavior change. Curiosity about personal dietary data and supporting broader research goals were the most common reasons for enrolling (79.8%, respectively). Notably, the foods participants flagged as missing or unexpected clustered among herbs, spices, seafood, and processed items, the foods most easily forgotten in self-report and least readily assigned from sequence. Overall, our results suggest that genomic dietary data can be returned in a form participants find understandable and engaging, and that curiosity alone can motivate participation without payment.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.