Text mining in support of psychiatric research: overcoming fragmented knowledge (PsyMine)

Mental disorders belong to the top contributors of the contemporary burden of disease worldwide. Currently, however, a researcher or clinician seeking to better understand the causal underpinnings of a disorder encounters an overwhelming amount of literature, which is distributed across several fields of study (e.g. Molecular Genetics, Neurobiology, Psychiatry, and Sociology). Furthermore – and quite paradoxically – the rapid increase in the number of scientific articles published each year prevents scientists from gaining a complete picture of disorders. That is, distinct research communities studying the same disorder lay their own claim to causes and drill down into them, rather than developing an integrative understanding of the disorder.

In the proposed project, we develop the prototype of a text mining application that is specifically designed to overcome this fragmentation of knowledge. The application’s main output will be a database of cause-disorder-relationships detected in the scientific literature across disciplinary boundaries. This database will provide researchers with the targeted, specific results they expect from conducting a literature search, but due to the increased computational capacity of the automated application, it will provide unprecedented breadth. The application has the potential to transform research in the mental health field. Scholars will be capable of asking integrative questions on the nature of mental disorders and benefiting from the work of scientists across the spectrum of disciplines. The project consists of three work packages, 1) the development of the required vocabulary, 2) the development of the relation-extraction application, and 3) the creation of a database of the extracted cause-disorder-relations, publicly accessible via a web-interface.