Method builds detailed knowledge graphs from cancer clinical notes
Ontology-Based Post-Coordinated Expression Assembly: from Clinical Narrative to Queryable Knowledge Graphs
Computation and Language
Summary
Clinical notes about cancer often contain important details written as regular sentences, which makes it hard for computers to search them easily. The authors created a method called UNIGRAPHE that turns these sentences into structured knowledge graphs by linking medical terms according to clear rules. This helps computers understand not just what terms appear, but how they relate, making it easier to search and analyze medical records. Their method was tested on real oncology documents and successfully converted thousands of annotations into meaningful, searchable expressions.
What this means in practice
- •For hospital data teams: Automatically transform free-text oncology reports into structured graphs for precise patient cohort identification.
- •For health it developers: Build clinical search tools that leverage relations between medical concepts rather than isolated terms in notes.
Tested on one dataset.
Authors
Johann Pignat, Christophe Gaudet-Blavignac, Julien Ehrsam, Jean-Philippe Goldman, Cyrille Duret, Phil Cheng, Mina Bjelogrlic, Michel Cuendet, Christian Lovis, Olivier Michielin
Abstract
Clinical oncology documentation records decisive information in free text, where it remains readable but not retrievable at scale, leaving cohort identification dependent on incomplete diagnostic codes or manual review. The current standard, coding individual terms, captures what a document mentions, not what it asserts: meaning lies in how those terms relate to one another. We present a method for deriving queryable knowledge graphs from clinical narratives, conforming to the Extended Grammar of SNOMED CT across multiple ontologies. The method specifies how entities should be annotated and related. It defines five relation classes, each bound to a construction rule that assembles the entities and relations into post-coordinated expressions. A deterministic implementation, UNIGRAPHE, applies those rules to construct knowledge graphs from annotated text. Applied to expert-annotated real-world oncology documentation, UNIGRAPHE reconstructed 5,299 annotations into 2,196 post-coordinated expressions and shows that relation structure is the missing piece to make clinical narrative formally queryable.