Man holding stomach
Image source: istock.com/Prostock-Studio
The UK Health Security Agency (UKHSA) is exploring the role that AI could play in helping scientists to detect and investigate outbreaks of foodborne illness.
It has published a study of its evaluations of different types of AI for their ability to detect and classify text in online restaurant reviews – screening for key terms such as diarrhoea, vomiting and abdominal pain – and mentions of specific types of food.
Over 3,000 reviews were manually annotated by epidemiologists after being collected and filtered.
The types of AI, including large language models (LLMs), could one day be used to identify and potentially target investigations into foodborne illness outbreaks.
UKHSA said that previous research has explored similar approaches but that its study is more comprehensive, examining a more detailed list of terms and language patterns.
Data challenges
The research has also highlighted challenges that have to be addressed before widespread implementation, particularly around data access and quality.
While it is possible to use the approach to gather general information on the type of food people have eaten and which may be linked to illness, determining which specific ingredients or other factors that may be linked is difficult.
Variations in spelling and the use of slang were also identified as potential challenges, as well as people misattributing their illness to a given meal.
Professor Steven Riley, chief data officer at UKHSA said: “We are constantly looking for new and effective ways to enhance our disease surveillance.
“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.
“Further work is needed before we adopt these methods into our routine approach to tackling foodborne illness outbreaks.”
Patient experience
UKHSA has reported on two other projects exploring the use of AI, one of which is using LLMs to accelerate the analysis of qualitative survey data on patients’ lived experiences in the healthcare system.
This drew on responses from the Positive Voices 2022 survey of people living with HIV to identify key themes. It is currently undergoing human evaluation.
The other project has involved the use of LLMs on UKHSA’s computing clusters to automatically detect potential conflicts between public health guidance recommendations. The system enables users to upload a piece of guidance in development and automatically retrieves existing, relevant sections of guidance, then flags any potential conflicts.
This is currently being user tested internally, and UKHSA said it shows promising early results with over 90% accuracy rates.
It added that the tool could help to ensure that public health messaging is clear and consistent, especially during fast moving health emergencies.
Enhancing protection
Dr Nick Watkins, the agency’s chief data scientist, said: “These projects demonstrate how, alongside human expertise, AI can enhance public health protection.
“As we continue to develop and refine these systems, we maintain a careful balance between embracing innovation and ensuring robust validation of AI outputs. This approach helps us harness AI's potential while maintaining the high standards expected of a national public health agency.”