LinkMed: Entity Recognition and Relation Extraction from Clinical Notes in Spanish

  • Carlos Muñoz-Castro
  • , Andrés Carvallo
  • , Matías Rojas
  • , Claudio Aracena
  • , Rodrigo Guerra
  • , Benjamín Pizarro
  • , Jocelyn Dunstan

Research output: Contribution to journalConference articlepeer-review

Abstract

Relation extraction is an essential component of Natural Language Processing (NLP) and significantly influences information retrieval and structured information extraction. Within clinical notes, the task is needed to establish connections among illnesses, therapies, indications, and other medical concepts. Motivated by the above, in this work, we propose a two-step model approach for entity linking; in the first step, we solve entity recognition, and in the second, a relation classification approach. We evaluated our approach in a Spanish corpus of the TESTLINK challenge in IberLEF2023 (Iberian Languages Evaluation Forum), comprising 81 clinical notes to train and 80 clinical notes to test. Our results show competitive performance with a precision of 0.47, recall of 0.43, and F1-score of 0.45, presenting an effective strategy for relation extraction from clinical notes in Spanish.

Original languageEnglish
JournalCEUR Workshop Proceedings
Volume3496
StatePublished - 2023
Externally publishedYes
Event2023 Iberian Languages Evaluation Forum, IberLEF 2023 - Jaen, Spain
Duration: 26 Sep 2023 → …

Keywords

  • Clinical Text
  • Link prediction
  • Named Entity Recognition
  • Natural Language Processing

Fingerprint

Dive into the research topics of 'LinkMed: Entity Recognition and Relation Extraction from Clinical Notes in Spanish'. Together they form a unique fingerprint.

Cite this