2021/10/19 by Ali Can Kocabiyikoglu, François Portet, Kocabiyikoglu, Ali Can +5
Biochemistry, Genetics and Molecular Biology · Computer Science · #Biomedical Text Mining and Ontologies #Computation and Language (cs.CL) #FOS: Computer and information sciences #Machine Learning in Healthcare #Topic Modeling
paper · pdf · doi:10.48550/arxiv.2110.10213
openalex publication_date 2021/10/19 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Drug prescriptions are essential information that must be encoded in\nelectronic medical records. However, much of this information is hidden within\nfree-text reports. This is why the medication extraction task has emerged. To\ndate, most of the research effort has focused on small amount of data and has\nonly recently considered deep learning methods. In this paper, we present an\nindependent and comprehensive evaluation of state-of-the-art neural\narchitectures on the I2B2 medical prescription extraction task both in the\nsupervised and semi-supervised settings. The study shows the very competitive\nperformance of simple DNN models on the task as well as the high interest of\npre-trained models. Adapting the latter models on the I2B2 dataset enables to\npush medication extraction performances above the state-of-the-art. Finally,\nthe study also confirms that semi-supervised techniques are promising to\nleverage large amounts of unlabeled data in particular in low resource setting\nwhen labeled data is too costly to acquire.\n