NHS deploys ClinQueryAgent for natural language health data queries

ClinQueryAgent is a novel conversational agent architecture designed for population health management, developed by researchers including Joseph S. Boyle and colleagues. The system bridges the gap between clinical natural language and database queries by leveraging powerful cloud-based language models. Critically, the architecture ensures that patient data never leaves the secure NHS environment—cloud models are used as a service but data is anonymized and queries are handled locally. To maintain accuracy over longer interactions, a dedicated sub-agent handles information retrieval, preventing the context rot that plagues many conversational systems.

In a real-world deployment across 15 healthcare practices in the UK's National Health Service, 128 staff—including both analysts and clinicians—used ClinQueryAgent to query patient records covering 148,319 individuals. The results showed that users could easily generate actionable information from health records using natural language, without requiring any programming expertise. The system was evaluated on a constructed dataset and during a beta-testing phase. A public demo is available at the link in the paper, and the work has been submitted to ACL Systems Demonstrations.