One Good Choice (1GC)

One Good Choice (1GC) combines large language models, machine learning, and curated data sources to predict caloric density and a Food Compass Score from a simple text description, offering personalized, healthier alternatives.

Unlike traditional calorie trackers, 1GC is a decision-making platform that proactively connects with individuals, educating and offering personalized recommendations to inform dietary decisions, one meal at a time.

Data Sources

Food Compass 2.0 — Comprehensive nutrition profiling system scoring foods 1-100 (Tufts University)
USDA FoodData Central — Calorie content and nutritional metadata

AI Pipeline (7 steps)

Baseline profiling: predict calories and Food Compass Score (FCS)
LLM generates 15 distinct healthier alternatives
Each alternative evaluated in parallel through nutritional prediction pipeline
Suggestions filtered by goals: improved FCS, lower calories, or both
Recipe search from trusted nutrition sites
LLM ranks recipe URLs by semantic relevance
Results displayed in clean interface

Models

Food Compass Score: Feedforward NN (768→256→128→64→1) using 768-dim sentence-transformer embeddings
Calorie Estimation: TF-IDF + cooking-method multipliers + 1024-dim embeddings → regression

Paper: arXiv:2512.11836

Data Sources

AI Pipeline (7 steps)

Models

Related

Platforms Used