Data Disambiguation Assistant

Reducing manual review through automated data cleanup in the Tuberculosis Screening Programme

  • Publicly sourced

The challenge

The UK's Tuberculosis Screening Programme invites new migrants for testing based on their country of birth, using data from GP registration records. However, these records often include incomplete, misspelled, or ambiguous place names, resulting in messy data that requires extensive manual review. On average, around 40,000 records must be reviewed each year to ensure accurate screening.

The solution

The team used AI to automatically match inconsistent place name data to countries, significantly reducing the need for manual checks. The solution is now fully integrated into the Tuberculosis Screening Programme team's regular workflow.

NOTE: An Algorithmic Transparency Recording Standard (ATRS) record is currently being prepared.

The results

  • Achieved 90% accuracy in correctly identifying country of birth from inconsistent data.
  • Reduced manual review workload by 85%, from 40,000 to 6,000 records per year.

Details

Organisation name
Department of Health and Social Care (DHSC) / UK Health Security Agency (UKHSA)
Government body
UK Government
User group
Wider public sector
Use case type
Specific
Type of technology
Machine Learning
Phase
Live
Impact
Improved efficiency / Time savings

Links

Get in touch

Emailai-knowledge-hub@dsit.gov.ukto:

  • find out how to use or collaborate on this tool
  • review content for this tool
  • talk to us about your own tool

Content created: 10 July 2025