CamemBERT-bio: Leveraging Continual Pre-training for Cost-Effective Models on French Biomedical Data
This work addresses the problem of extracting information from unstructured French biomedical documents for clinical research, though it is incremental as it builds on existing pre-training methods.
The authors tackled the inefficiency of existing French language models on biomedical data by introducing CamemBERT-bio, a model derived from continual pre-training on a new French biomedical dataset, which achieved an average improvement of 2.54 F1-score points on biomedical named entity recognition tasks.
Clinical data in hospitals are increasingly accessible for research through clinical data warehouses. However these documents are unstructured and it is therefore necessary to extract information from medical reports to conduct clinical studies. Transfer learning with BERT-like models such as CamemBERT has allowed major advances for French, especially for named entity recognition. However, these models are trained for plain language and are less efficient on biomedical data. Addressing this gap, we introduce CamemBERT-bio, a dedicated French biomedical model derived from a new public French biomedical dataset. Through continual pre-training of the original CamemBERT, CamemBERT-bio achieves an improvement of 2.54 points of F1-score on average across various biomedical named entity recognition tasks, reinforcing the potential of continual pre-training as an equally proficient yet less computationally intensive alternative to training from scratch. Additionally, we highlight the importance of using a standard evaluation protocol that provides a clear view of the current state-of-the-art for French biomedical models.