Food Additive Lens is a real, free iOS app (available on the App Store for iPhone 14 and newer, with a desktop/Mac version) that uses on-device AI to scan food ingredient labels and provide clear, science-based explanations of additives.
Key Details
- Developer: Yihang Feng ’25 (dual PhD in Nutritional Sciences and master’s in Computer Science at the University of Connecticut). He built it during a summer research assistantship at the Institute for the Advancement of Food and Nutrition Sciences (IAFNS), with input from advisors Yangchao Luo (Nutritional Sciences) and Song Han (Computer Science). Yi Wang has been involved in development and is taking over future work.
- How it works: Users photograph an ingredient list. A three-agent AI system then:
- Categorises the food (trained on >10,000 items from USDA’s Global Branded Food Products Database).
- Identifies additives (drawing from >4,000 FDA-approved ones).
- Generates plain-language explanations of what they are and their role in food.
- Deeper regulatory/technical details are available for professionals. It uses on-device AI (including a quantised Meta Llama 3.2 3B model with Retrieval-Augmented Generation) for speed, privacy, and offline functionality—no internet required after setup.
- Sources: Relies on authoritative data like the FDA’s Substances Added to Foods Database, Code of Federal Regulations, etc., to reduce misinformation and hallucinations.
- Publication: A paper detailing the app’s development, architecture, performance (e.g., classifier accuracy, additive identification F1-score), and consumer education focus was published in Digital Discovery.
The app addresses a common pain point: reliable additive info exists in scientific/regulatory sources, but is hard for everyday shoppers to access quickly. Feedback from testers helped refine the UI. Future goals include more personalised recommendations based on individual health/dietary needs.
You can download it here: Food Additive Lens on the App Store.
