NAMED ENTITY RECOGNITION FOR MEDICAL DATA EXTRACTION USING BIOBERT

Abstract

Technological advancements have caused widespread shifts in the medical industry. A vast quantity of information may be found in the medical literature publications released by researchers. Natural language processing innovations have made it simple to extract information on drugs, illnesses, symptoms, routes doses, species, and routes of administration from a documented source. This proposed research is used to identify the named entities from the medical literature. The BioBERT model is used to train the corpus that has been annotated. The proposed framework can outperform many state-of-the-art baselines and provide state-of-the-art results for BioNER. When compared to the existing model, the accuracy provided by the proposed system is satisfactory. The BioBERT is used for the extraction of medical entities like drugs, chemicals, genes, etc and to train the corpus that has been annotated. The BioBERT has in-built data of the medical entities, that will identify all the medical-related data or entities from the given statement. More entities can be found by this method than by the current standard model.When compared to the existing model, the accuracy provided by the trained version is satisfactory.The proposed system comes with a GUI for users to type the clinical words or components for analysis.

Keywords

• natural language processing
• named entity recognition
• scispacy
• biobert
• science
• medicine data