Learning embeddings from free-text triage notes using pretrained transformer models

Arnaud, Émilien; Elbattah, Mahmoud; Gignon, Maxime; Dequen, Gilles

Abstract

The advent of transformer models has allowed for tremendous progress in the Natural Language Processing (NLP) domain. Pretrained transformers could successfully deliver the state-of-the-art performance in a myriad of NLP tasks. This study presents an application of transformers to learn contextual embeddings from free-text triage notes, widely recorded at the emergency department. A large-scale retrospective cohort of triage notes of more than 260K records was provided by the University Hospital of Amiens-Picardy in France. We utilize a set of Bidirectional Encoder Representations from Transformers (BERT) for the French language. The quality of embeddings is empirically examined based on a set of clustering models. In this regard, we provide a comparative analysis of popular models including CamemBERT, FlauBERT, and mBART. The study could be generally regarded as an addition to the ongoing contributions of applying the BERT approach in the healthcare context.

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 15th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2022)
• Start Date: Feb 9, 2022
• End Date: Feb 11, 2022
• Acceptance Date: Dec 15, 2021
• Publication Date: 2022
• Deposit Date: May 11, 2022
• Volume: 5
• Pages: 835-841
• Book Title: Proceedings of the 15th International Joint Conference on Biomedical Engineering Systems and Technologies
• ISBN: 9789897585524
• DOI: https://doi.org/10.5220/0011012800003123
• Public URL: https://uwe-repository.worktribe.com/output/9187291