Large Language Models for Temporal Relation Extraction from French Clinical Narratives — Prompt-Chaining Approach
OVERVIEWPrompting large language models with a minimal set of in-context examples can achieve strong performance for temporal relation extraction (TRE) and normalization in low-resource settings with limited domain-specific annotations. A prompt-chaining approach leveraging open-weight, on-premises models offers a practical pathway toward deployment in real-world clinical workflows, particularly in French-language healthcare systems. This work addresses a critical bottleneck in clinical NLP: the need for reliable information extraction from unstructured clinical narratives in non-English languages and resource-constrained settings. By demonstrating that LLMs can perform sophisticated temporal reasoning with minimal training data, the research enables broader application of AI to clinical documentation analysis across diverse language contexts. The approach can be further extended to extract key clinical variables such as age at diagnosis or age at onset—particularly valuable in rare disease contexts where data scarcity limits traditional machine learning approaches. This framework supports the development of multilingual, low-resource clinical AI systems that can operate on-premises without reliance on proprietary cloud infrastructure. ---
Subscribe to read full wiki analysis
Full wiki content — including source provenance, contradiction tracking, related pages, and analyst notes — is available to Subscriber members.
View Plans & SubscribeNew users get a 7-day trial — no credit card required.