RNN embeddings for identifying difficult to understand medical words - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2019

RNN embeddings for identifying difficult to understand medical words

Résumé

Patients and their families often require a better understanding of medical information provided by doctors. We currently address this issue by improving the identification of difficult to understand medical words. We introduce novel embeddings received from RNN - FrnnMUTE (French RNN Medical Understandability Text Embeddings) which allow to reach up to 87.0 F1 score in identification of difficult words. We also note that adding pre-trained FastText word embeddings to the feature set substantially improves the performance of the model which classifies words ac- cording to their difficulty. We study the generalizability of different models through three cross-validation scenarios which allow testing classifiers in real-world conditions: understanding of medical words by new users, and classification of new unseen words by the automatic models. The RNN - FrnnMUTE embeddings and the categorization code are being made available for the research.
Fichier non déposé

Dates et versions

hal-02371219 , version 1 (19-11-2019)

Identifiants

  • HAL Id : hal-02371219 , version 1

Citer

Hanna Pylieva, Artem Chernodub, Natalia Grabar, Thierry Hamon. RNN embeddings for identifying difficult to understand medical words. ACL Workshop on Biomedical Natural Language Processing, Aug 2019, Florence, Italy. ⟨hal-02371219⟩
70 Consultations
0 Téléchargements

Partager

Gmail Facebook X LinkedIn More