TRExt: TRE Text Analytics

Unstructured text, such as clinical notes and reports, contains valuable research insights but is difficult to analyse within current Trusted Research Environment (TRE) infrastructures. Many environments lack shared tools for text analytics, particularly where data cannot be moved or centralised. This limits cross-site research using natural language processing methods.

TRExt will integrate existing text extraction and structuring tools into a reusable pipeline compatible with the Five Safes TES infrastructure for federation. Rather than exposing raw text, the pipeline will transform unstructured data into analysable representations that can be queried across environments. Demonstrations will use anonymised and synthetic datasets.

The project will help TRE operators and researchers understand how text analytics can be delivered responsibly and at scale. Its outputs will support more consistent adoption of text analysis capabilities across the TRE ecosystem.

TRExt will work closely with the SAIL Databank Public and Patient Advisory Group to shape its Public Involvement and Engagement strategy as a living plan. Public workshops will explore trust, concerns, and expectations around text analytics, with anonymised findings informing both technical and governance decisions.

By the end of the project, TRExt will:

• Deliver a pipeline for enabling federated analysis of unstructured text in TREs
• Demonstrate safe text analytics using anonymised and synthetic data
• Produce public-informed guidance on acceptable text data use
• Support wider adoption of text analytics capabilities

Project information

Lead organisation: University of Nottingham
Principal investigator: Dr Tim Beck
Project duration: 12 months
Project partners: King’s College London, Swansea University
Funding provided: £310,975
Primary contact email: tim.beck@nottingham.ac.uk