In the Pharmaceutical and Drug Safety domain, a lot of money and time is invested on drug discovery and health treatment every year. Adverse drug reactions (ADRs) cause a significant number of deaths worldwide and billions of dollars are spent yearly to treat people who had an ADR from a prescribed drug. Anthi Symeonidou with Prof. Paul Groth and Viachaslau Sazonau have applied a machine learning approach which identifies ADRs and achieves substantially better results than previous approaches.
Adverse drug reactions
Transfer Learning for Biomedical Named Entity Recognition with BioBERT the title of the paper. In this project a transfer learning approach is applied to ADR recognition on three biomedical datasets and being compared with traditional approaches, such as dictionary, conditional random fields and BiLSTM. The main contribution is empirical and shows that transfer learning method based on BioBERT can achieve considerably higher performance in recognizing ADRs than traditional methods. The results suggest that transfer learning based on transformer architecture is a promising approach to addressing the lack of training data in biomedical Information Extraction.
SEMANTiCS conference is a leading European conference on Semantic Technologies and AI, where researchers, industry experts and business leaders can develop a thorough understanding of trends and application scenarios in the fields of Machine Learning, Data Science, Linked Data and Natural Language Processing.