Medical NER for French Language
Résumé
Medical practitioners and other professionals in the healthcare sector face a considerable amount of documents to manage, in particular medical reports. These documents must be indexed and organized so that it is possible to rapidly retrieve essential information concerning the patients, to comprehend their care pathway and to chart their clinical profile. In order to convert unstructured text into an annotated ontology leveraging a better understanding of a specific domain, Named Entity Recognition (NER), one of the most important tasks in Natural Language Processing (NLP), is a particularly promising solution. Several implementations of medical NER can already be found in the literature, however these models are typically developed for the English language. In this paper we present the design and the implementation of a custom NER model to identify named entities for medical reports written in the French language. The used collection consists of printed reports from various origins which have been processed by Optical Character Recognition (OCR) technology. Considering the limited training data, the obtained results are satisfactory and confirm the validity of the proposed approach