2018/02/14 by Shashank Gupta, Gupta, Shashank, Manish Gupta +9
Biochemistry, Genetics and Molecular Biology · Computer Science · Pharmacology, Toxicology and Pharmaceutics · #Biomedical Text Mining and Ontologies #Computation and Language (cs.CL) #Computational Drug Discovery Methods #FOS: Computer and information sciences #Information Retrieval (cs.IR) #Pharmacovigilance and Adverse Drug Reactions
paper · pdf · doi:10.48550/arxiv.1802.05130
openalex publication_date 2018/02/14 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Adverse drug reactions (ADRs) are one of the leading causes of mortality in\nhealth care. Current ADR surveillance systems are often associated with a\nsubstantial time lag before such events are officially published. On the other\nhand, online social media such as Twitter contain information about ADR events\nin real-time, much before any official reporting. Current state-of-the-art in\nADR mention extraction uses Recurrent Neural Networks (RNN), which typically\nneed large labeled corpora. Towards this end, we propose a multi-task learning\nbased method which can utilize a similar auxiliary task (adverse drug event\ndetection) to enhance the performance of the main task, i.e., ADR extraction.\nFurthermore, in the absence of auxiliary task dataset, we propose a novel joint\nmulti-task learning method to automatically generate weak supervision dataset\nfor the auxiliary task when a large pool of unlabeled tweets is available.\nExperiments with 0.48M tweets show that the proposed approach outperforms the\nstate-of-the-art methods for the ADR mention extraction task by 7.2% in terms\nof F1 score.\n