Deep Learning for Extracting Biomedical Entities from COVID-19 Dataset: A Case Study

Biomedical entity extraction is most primitive task of biomedical text mining. There are a lot of tools for extracting general biomedical entities such as: disease, treatment, proteins, chemical, etc. However with the emergence of COVID-19, a lot of textual data was generated and the existing tools were not sufficient for extracting meaningful entities for COVID. Many new entities were identified for COVID that provided a rich source of information for the disease. In order to provide a machine learning platform for extracting entities from COVID-related text data, a benchmark dataset—CORD-NER was developed. A lot of deep learning models were also applied for this task. This chapter presents the architecture and experiments of two popular deep learning data models: LSTM and BERT for COVID entity extraction task.