The United Kingdom Health Security Agency (UKHSA) is investigating how artificial intelligence (AI) might enhance detection of foodborne illness outbreaks by analyzing online restaurant reviews, according to WiredGov. A recent study evaluated various AI large language models for their effectiveness in identifying text related to gastrointestinal symptoms and food types in customer feedback.
Foodborne gastrointestinal illness, typically causing vomiting and diarrhea, affects millions of UK residents annually. However, most cases go undiagnosed through traditional medical channels, creating significant gaps in surveillance data.
UKHSA technology experts and scientists assessed AI systems’ ability to scan thousands of online reviews for indicators of gastrointestinal illness, including specific symptoms such as diarrhea, vomiting, and abdominal pain. The technology also tracked information about the types of food consumed by reviewers who reported illness.
Researchers believe this approach could eventually become a standard surveillance tool, providing valuable insights into unreported illness rates while offering critical clues about potential outbreak sources. The method could supplement existing detection systems by capturing data from individuals who experience symptoms but don’t seek medical care.
Despite promising results, the study identified several challenges that would need to be addressed before implementation. Key obstacles include securing access to real-time review data and distinguishing between general food categories and specific ingredients that might be responsible for illnesses. Additional complications include variations in spelling, use of slang terms, and customers incorrectly attributing their illness to particular meals.
“We are constantly looking for new and effective ways to enhance our disease surveillance,” said Professor Steven Riley, Chief Data Officer at UKHSA. “Using AI in this way could soon help us identify the likely source of more foodborne illness outbreaks, in combination with traditional epidemiological methods, to prevent more people becoming sick.”
The UKHSA study expands on previous research in this area by examining a more comprehensive list of terms and language patterns that could help identify illness outbreaks. For the analysis, epidemiologists manually annotated over three thousand reviews after filtering them for gastrointestinal illness keywords. The researchers specifically focused on symptoms like diarrhea and vomiting while excluding less specific indicators such as headaches, fever, or respiratory symptoms.
This initiative represents part of UKHSA’s broader evaluation of artificial intelligence applications in public health surveillance and response systems. Professor Riley noted that further refinement would be necessary before these methods could be incorporated into routine outbreak investigation protocols.
Commenting on this article, the nation’s leading food poisoning lawyer said, “Technology like artificial intelligence holds a lot of promise to aid in the fight against foodborne illness. Over time, outbreaks may be able to be identified faster, which could reduce the amount of illnesses and overall risk to the public.”
