A UK healthtech company needed to process thousands of unstructured patient records for clinical trial matching. We built an NLP system that extracts medical entities, maps them to standardised ontologies, and matches patients to eligible trials automatically.
“We were watching patients miss their trial windows simply because we couldn’t process records fast enough. Every day of delay was a day a patient might lose access to a treatment that could change their outcome.”
— Clinical Operations Lead, UK Healthtech CompanyMapped document types and data flows across 3 NHS trust partners. Identified 200+ medical entities critical for trial eligibility, catalogued format variations, and established a ground-truth annotation dataset with clinical domain experts.
Built a custom biomedical Named Entity Recognition (NER) model trained on UK clinical text, including GP correspondence and NHS discharge summaries. Mapped all extractions to SNOMED CT and ICD-10 ontologies for standardised, interoperable patient profiles.
Developed a semantic matching algorithm that compares structured patient profiles against trial eligibility criteria in real time. Each match is accompanied by a confidence score, enabling clinicians to prioritise high-probability candidates while reviewing borderline cases.
Deployed within NHS-approved cloud infrastructure with full IG Toolkit compliance. Implemented role-based access controls, end-to-end audit logging, and data residency guarantees to meet the highest standards of patient data protection.
Book a free AI readiness call. We'll discuss your challenges and outline a clear path forward.