Using concepts in literature-based discovery: Simulating Swanson's Raynaud-fish oil and migraine-magnesium discoveries

Literature‐based discovery has resulted in new knowledge. In the biomedical context, Don R. Swanson has generated several literature‐based hypotheses that have been corroborated experimentally and clinically. In this paper, we propose a two‐step model of the discovery process in which hypotheses are generated and subsequently tested. We have implemented this model in a Natural Language Processing system that uses biomedical Unified Medical Language System (UMLS) concepts as its unit of analysis. We use the semantic information that is provided with these concepts as a powerful filter to successfully simulate Swanson's discoveries of connecting Raynaud's disease with fish oil and migraine with a magnesium deficiency.