TRExt is developing a secure way to turn written medical records into research-ready data, enabling researchers to safely analyse information across multiple Trusted Research Environments (TREs) while protecting privacy. That could open new avenues of research across a wide range of health conditions, ultimately benefiting patients and the public.
Every day, clinicians write detailed notes about their patients, including observations, diagnoses, treatment plans, and outcomes, in unstructured, free-flowing language. This information is clinically rich and holds considerable research potential, but it has remained largely out of reach for researchers working within TREs. There is a technical barrier: converting free-text into the standard data formats that TREs use for analysis, in a way that enables it to be queried across multiple TREs simultaneously. This has not been possible until now, leaving an enormous body of clinical evidence recorded in letters, discharge summaries, and reports currently unavailable to researchers.
By solving this problem, TRExt (TRE Text Analytics) aims to unleash the detail contained in clinical free-text, such as nuanced symptom descriptions, treatment responses, clinical reasoning, and the social context of people’s everyday lives that is often absent from the structured, coded records that researchers can currently access within TREs.
TRExt is led by the University of Nottingham in partnership with King’s College London and Swansea University, in collaboration with Hyper Unison, a company that specialises in data model mapping, and with a lead public representative embedded in the project management team.
The project explores two next-generation capabilities for TREs. The first is natural language processing (NLP), which uses computer-based methods to extract meaningful information from free-text. The second is data standardisation, which converts extracted information into formats compatible with international research data standards, enabling standardised access to data across multiple TREs for federated research.
What TRExt aims to achieve
By the end of the project, the TRExt team aims to demonstrate that clinical free-text can be made research-ready within TREs. Their goals include:
- Embedding public involvement – engaging the public from the very beginning of the project to ensure that the approach is trustworthy, transparent, and shaped by the people it is ultimately designed to benefit
- Transforming free-text for use in TREs – building a modular, reusable pipeline that converts clinical free-text into structured data conforming to the OMOP Common Data Model, an international standard for health research data
- Demonstrate model applicability – validating the pipeline using both a publicly available dataset and real NHS clinical text, to demonstrate applicability in live healthcare settings
- Bridging research standards – extending the pipeline’s reach by enabling the same structured data to be queried using the CDISC Study Data Tabulation Model (SDTM), a second international standard, used in clinical trials research
- Supporting federation – integrating the pipeline with the Five Safes TES federated analytics framework that allows researchers to run analyses across multiple TREs
Progress so far
Within the first six months of delivery, the TRExt team has:
- Built and integrated the core NLP pipeline
- Validated the pipeline using the publicly available MIMIC-IV health record dataset, demonstrating end-to-end processing of a large clinical text corpus into a research-ready OMOP database
- Extended the pipeline to merge NLP-derived data with existing structured health data, producing unified, person-level OMOP records
- Demonstrated cross-standard compatibility by enabling OMOP data to be queried using the CDISC SDTM standard without duplicating the underlying data
- Integrated the pipeline with the Five Safes TES federated analytics framework, demonstrating for the first time that data originally held in unstructured text can be queried securely across TREs
- Run public involvement and engagement workshops, with findings shaping the technical design of the pipeline and supporting the co-production of resources to communicate complex processes to a public audience, including a co-designed infographic
In the second half of the project, the team will validate the pipeline using real NHS clinical text, evaluate quality and pipeline performance, and publish open-source tools and guidance to support adoption across the TRE community.
Relevance to DARE UK vision
TRExt directly supports the DARE UK vision of expanding what is possible within TREs by tackling one of the most significant remaining barriers to their use: the exclusion of unstructured text from federated research. By delivering open-source, reusable infrastructure and demonstrating its application with real NHS data, the project will generate practical evidence and tools that TRE operators across the UK can adopt, extending the reach and depth of secure health data research for public benefit.
About the DARE UK Next-Gen Catalysts
The DARE UK Next-Gen Catalysts are early-stage, exploratory projects developing and testing the next generation of capabilities for Trusted Research Environments (TREs) across the UK. Bringing together researchers, technology specialists, public contributors and other partners, these projects explore bold new ideas that could expand how sensitive data is accessed, analysed and governed while maintaining privacy, security and public trust. Their role is to generate evidence, prototypes and practical learning that can help shape the future of secure, trustworthy data research for public benefit.
