MétaCan
Menu
Back to cohort
Record W3172825397 · doi:10.1093/cdn/nzab052_004

Rx Food App: A Proof-of-Concept Study of an Image-Based Dietary Assessment Mobile Application

2021· article· en· W3172825397 on OpenAlex
Katherine Jefferson, Elizabeth Choi, Derrick Lichti, Jeffrey Alfonsi, Barkha P. Patel, Jill Hamilton, JoAnne Arcand

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.

affAt least one author lists a Canadian institution in the pinned OpenAlex snapshot.

Bibliographic record

VenueCurrent Developments in Nutrition · 2021
Typearticle
Languageen
FieldMedicine
TopicNutritional Studies and Diet
Canadian institutionsSickKids FoundationWestern UniversityOntario Tech University
Fundersnot available
KeywordsFood groupNutrientFood composition dataAdded sugarFood scienceMathematicsSugarLimits of agreementAnimal scienceMedicineChemistryBiologyEnvironmental healthNuclear medicine

Abstract

fetched live from OpenAlex

To determine if Rx Food, an image-based dietary assessment app powered by artificial intelligence, can derive comparable nutritional composition estimates compared to calculated methods. Sub-group analyses assessed differences between composite (i.e., multiple ingredients) and single item foods. Food items were selected for testing based on their frequency of consumption among patients attending a weight management clinic. Food photos were uploaded, and serving sizes entered, into the app which generated estimated nutrient data. The nutritional composition of foods was also analyzed with ESHA Food Processor software. Nutrient estimates between the methods were compared using paired t-tests, Pearson correlation coefficients, and Bland-Altman plots for energy, carbohydrates, protein, total fat, fibre, total sugar and sodium. Thirty-nine food items were analyzed [n = 10 (27%) composite items and n = 29 (73%) single item foods]. There were no statistically significant differences in the mean differences in estimates from Rx Food and calculated values for all nutrients: −4.3 ± 29.2 kcal for energy, −0.4 ± 2.6 g for carbohydrates, −0.1 ± 1.9 g for protein, −0.3 ± 1.7 g for fat, −0.2 ± 2.3 g for fibre, 0.01 ± 1.4 g for sugar, and −33 ± 135 mg for sodium. Among all food items, a strong, significant correlation (r > 0.80; P < 0.05) was observed for all nutrients except fibre (r = 0.552; P < 0.001). In the Bland-Altman plots for all foods, significant bias was found for fibre (r = 0.562; P < 0.001), fat (r = 0.562; P = 0.025), and sodium (r = 0.359; P = 0.025), suggesting that Rx Food may underestimate nutrient composition at higher levels. Subgroup analyses of composite items showed significant strong correlations for energy, carbohydrates, protein, and sugar (r > 0.80; P < 0.05), significant moderate correlations (r = 0.60–0.79; P < 0.05) for fat and fibre, but not for sodium (r = 0.591; P = 0.072). Single item analysis showed significant correlations for all nutrients (r > 0.80; P < 0.05). This preliminary data shows that Rx Food has the potential to be an accurate, image-based, low burden tool to calculate nutrient composition of foods. These findings justify further research to determine the validity of Rx Food in its ability to generate accurate nutrient intake data as a dietary assessment tool. N/A.

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 imitation

Not 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.

metaresearch head score (Codex)0.000
metaresearch head score (Gemma)0.000
Version: codex-gemma-dda1882f352aValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Observational · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.457
Threshold uncertainty score0.625

Codex and Gemma teacher scores by category

CategoryCodexGemma
Metaresearch0.0000.000
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0000.000
Science and technology studies0.0000.000
Scholarly communication0.0000.000
Open science0.0000.000
Research integrity0.0000.000
Insufficient payload (model declined to judge)0.0000.000

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.

Opus teacher head0.036
GPT teacher head0.355
Teacher spread0.319 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it