University College London Researchers Launch Pioneering AI Tool to Quantify Health Risks of Nutrition Misinformation

UCL researchers launch Diet-MisRAT to identify and rank the potential for harm in nutrition misinformation, aiding regulators in curbing dangerous digital trends.

By: AXL Media

Published: Mar 27, 2026, 6:30 AM EDT

Source: Information for this report was sourced from University College London

Content Quantifying the Danger of Digital Dietary Advice

A breakthrough in public health technology has emerged from University College London as researchers unveil a sophisticated tool capable of measuring the specific harm potential of nutrition misinformation. Unlike traditional fact-checking mechanisms that provide binary true or false labels, the newly developed Diet-Nutrition Misinformation Risk Assessment Tool, known as Diet-MisRAT, evaluates content based on its likelihood to cause physical injury or death. This shift in methodology treats digital information as a medium for hazardous exposure, mirroring how health authorities monitor physical environmental threats to determine necessary intervention levels.

A Scalable Traffic Light System for Digital Safety

The framework operates as a rule-based analysis model that categorizes online health claims into green, amber, or red tiers based on a weighted risk score. According to lead developer Alex Ruani of the UCL Institute of Education, the tool was specifically designed to catch harmful content that often operates through selective framing rather than overt falsehoods. By identifying traits such as hazardous omissions and manipulative framing, the system can flag high-risk advice that might otherwise escape the notice of standard automated oversight until a medical crisis occurs.

Expert Validation and the Core Metrics of Risk

To ensure the model’s accuracy, the UCL team subjected Diet-MisRAT to five rigorous rounds of verification, incorporating feedback from nearly 60 specialists in the fields of dietetics, nutrition, and public health. This collaborative approach confirmed that the risk of harm is not just a product of the information itself, but also the context of the recipient and the prominence of the content. Professor Anastasia Kalea of the UCL Division of Medicine emphasized that including professional judgment was essential to ensure the tool’s output accurately reflects the nuances of medical risk in diverse populations.