FRASIMED: a Clinical French Annotated Resource Produced through Crosslingual BERT-Based Annotation Projection

The French Annotated Resource with Semantic Information for Medical Entities Detection (FRASIMED) contains 2'051 synthetic clinical cases in French, with 24'037 annotated entities. The dataset contains two subsets: CANTEMIST-FR: Originally from CANTEMIST (Miranda-Escalada et al. (2020)), it contains 1'301 oncological notes, with 15'978 annotations linked to an ICD-O-3.1 morphology code. Additionally, 15’457 of them are linked to a SNOMED-CT code. DISTEMIST-FR: Originally from DISTEMIST's training set (Miranda-Escalada et al. (2022)), it contains 750 clinical cases, with 8'059 annotations, with 5'132 of them linked to a SNOMED-CT code. Please, cite us: Zaghir, J., Bjelogrlic, M., Goldman, J.-P., Aananou, S., Gaudet-Blavignac, & Lovis, C. (2023). FRASIMED: a Clinical French Annotated Resource Produced through Crosslingual BERT-Based Annotation Projection. arXiv preprint http://arxiv.org/abs/2309.10770