Categorising ultra-processed foods in large-scale cohort studies: evidence from the Nurses’ Health Studies, the Health Professionals Follow-up Study, and the Growing Up Today Study
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
This manuscript details the strategy employed for categorising food items based on their processing levels into the four NOVA groups. Semi-quantitative food frequency questionnaires (FFQs) from the Nurses' Health Studies (NHS) I and II, the Health Professionals Follow-up Study (HPFS) and the Growing Up Today Studies (GUTS) I and II cohorts were used. The four-stage approach included: (i) the creation of a complete food list from the FFQs; (ii) assignment of food items to a NOVA group by three researchers; (iii) checking for consensus in categorisation and shortlisting discordant food items; (iv) discussions with experts and use of additional resources (research dieticians, cohort-specific documents, online grocery store scans) to guide the final categorisation of the short-listed items. At stage 1, 205 and 315 food items were compiled from the NHS and HPFS, and the GUTS FFQs, respectively. Over 70 % of food items from all cohorts were assigned to a NOVA group after stage 2. The remainder were shortlisted for further discussion (stage 3). After two rounds of reviews at stage 4, 95⋅6 % of food items (NHS + HPFS) and 90⋅7 % items (GUTS) were categorised. The remaining products were assigned to a non-ultra-processed food group (primary categorisation) and flagged for sensitivity analyses at which point they would be categorised as ultra-processed. Of all items in the food lists, 36⋅1 % in the NHS and HPFS cohorts and 43⋅5 % in the GUTS cohorts were identified as ultra-processed. Future work is needed to validate this approach. Documentation and discussions of alternative approaches for categorisation are encouraged.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.014 | 0.003 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it