LLM-based reconstruction of longitudinal clinical trajectories in chronic liver disease
A framework combining open-source large language models with constrained decoding to transform free-text radiology, pathology, and transplant assessment reports into structured clinical data. Applied to 22,493 reports from 835 patients, it reconstructs individual disease trajectories and supports research into early liver cancer detection.
Recommended citation: Paverd H, Gao Z, Mahani G, Fabre MA, Burge S, Hoare M, Crispin-Ortuzar M. LLM-based reconstruction of longitudinal clinical trajectories in chronic liver disease. npj Precision Oncology, 2026. https://doi.org/10.1038/s41698-026-01650-4.
Download Paper