aphp / edsnlp

Modular, fast NLP framework, compatible with Pytorch and spaCy, offering tailored support for French clinical notes.
https://aphp.github.io/edsnlp/
BSD 3-Clause "New" or "Revised" License
111 stars 29 forks source link

Include negation in tobacco and alcohol pipelines #287

Open marieverdoux opened 4 months ago

marieverdoux commented 4 months ago

With the current code, detailed status "ABSENT" is assigned only when negation is found after the entity, not before.

Example: If text is "Patient non fumeur", detailed_status is assigned to PRESENT. If text is "Fumeur: non" , detailed_status is assigned to ABSENT.

Feature suggestion: The negation patterns from negation pipelines should be included to the tobacco/alcohol pipelines so detailed_status is set to ABSENT in any negation case.