Text mining for neuroscience: A co-morbidity case study

Abstract

The vast amount of information available to researchers has accelerated the pace of scientific discovery but also challenges researchers to find novel ways to manage and extract information from large databases to drive this process. A solution to this problem is found in text mining, which refers to the process of employing automated algorithms to identify the salient features of databases and predict novel associations based on user-defined keywords. Neuroscience has witnessed an explosion in published research and assimilating information from the > 180 neuroscience journals and mapping these discoveries on to a century of published research becomes a daunting and time-consuming task for the researcher. More importantly, given the vastness of this literature, it is likely that the researcher will not detect converging or related variables. However, text mining provides an automated way for the researcher to mine neuroscience databases and this paper proposes to use this process to identify points of contact between two parallel yet well-established disorder-related literatures - schizophrenia and alcoholism. We mine these databases to identify common pathologies between these disease states to understand why these diseases are commonly comorbid with an eventual aim to identify rational novel treatment strategies. © Springer-Verlag Berlin Heidelberg 2013.