Generating hypotheses from medline: A new approach based on network analysis

Generating hypotheses from the medical literature is directed to discover associations between medical or biological entities which are not present in the text but can be inferred by the presence of other more explicit relationships. The purpose of hypothesis generation would reveal unknown relationships worthy of further investigation. Hidden relations are those that are unlikely to be found by examination of bibliographic citations or the use of indexing techniques and yet establish a relationship between topics that might be explored by scientific research. Within the discovery framework established by Swanson and Smalheiser, in the paper, we propose an experiment that uses network methods to discover hidden connections in the medical literature. Our experiment was conducted with the MEDLINE database.