UKHSA (UK Health Security Agency) uses the power of artificial intelligence (AI) to solve health security problems. Here are three examples of projects that show how to use state -of -the -art AI technology to protect and improve public health.
Use AI to understand the patient experience
Understanding the vivid experience of the patient is important for improving medical services. Traditionally, analyzing qualitative survey data was a lot of time and resource -intensive. We are currently using a large language model (LLM) to accelerate this process.
In a recent project that analyzes the response in a positive VOICES 2022 survey of people living in HIV, our AI system has successfully identified major themes in more than 1,600 responses to U = U (non -political = non -transformable) concepts. The AI -centered approach is currently being verified by human beings, but it shows the potential for AI to greatly accelerate the use of qualitative analysis in public health studies.
Using AI, it detects food -mediated diseases.
Food -like diseases have serious health burdens in the UK and affect millions of dollars every year, but in most cases, traditional surveillance remains not diagnosed. Our technical experts and scientists analyzed thousands of online reviews, including specific food types such as meat and fish, and evaluated various LLMs that can select major terms related to gastrointestinal symptoms such as vomiting and abdominal pain.
Previously, similar approaches were explored, but UKHSA’s research investigated a lot more detailed terms and language patterns, which are more comprehensive and helps to identify disease outbreaks. This study emphasized an important task to be solved before a wide range of implementation in relation to data access and quality.
Further work, this approach can potentially be a routine part of public health surveillance, and you can capture an example of providing important insights to possible sources as they slide and occur through the existing monitoring system.
Public health guidelines make it more consistent
Clear and consistent guidelines are essential during the health emergency. UKHSA maintains hundreds of guidelines for public and professionals, and consistency is important to reduce the risk of health risks. The manual review of the guideline document takes a lot of time and challenging. The recent project uses a safe LLM for UKHSA’s computing clusters to automatically detect potential conflicts between public health guidance. This system allows users to upload guidelines for development, automatically search existing related sections of UKHSA guidelines, and display potential conflicts. This is currently undergoing user tests internally and shows promising initial results at more than 90% accuracy in searching related texts and identifying inconsistencies in recommendations.
This tool can help keep public health messaging clear and consistent, especially during fast -moving health emergencies.
I am looking for it in the future
Since the AI is still in the early stages, you can imagine the potential benefits that can be brought to public health in the future.
Dr. Nick Watkins, Deputy Director of Data Science and Geographic Space, and UKHSA’s top data scientist said:
‘This project shows how AI can improve public health as well as human expertise. As we continue to develop and improve these systems, we maintain careful balance between accepting innovation and ensuring powerful verification of AI output. This approach helps to maintain high standards of national public health institutions by utilizing the potential of AI. ‘
