Understanding the complex connectivity structure of brain is a major challenge in neuroscience. Vast and ever expanding literature about neuronal connectivity between brain regions already exists in published research articles and databases. However, with the ever-expanding increase in published articles and repositories, it becomes difficult for a neuroscientist to engage with the breadth and depth of any given field within neuroscience. Natural Language Processing (NLP) techniques have been used to mine `Brain Region Connectivity' information from published articles in order to build a centralized connectivity resource helping neuroscience researchers to gain quick access to research findings. Manually curating and continuously updating such a resource involves significant time and effort.
This talk presents an application of supervised machine learning algorithms that perform shallow and deep linguistic analysis of text to automatically extract connectivity between brain region mentions. Proposed algorithms are evaluated using benchmark datasets collated from PubMed abstracts and a custom-built dataset of full-text PubMed articles annotated by a domain expert. Comparison with state-of-the-art methods including the popular transfer learning method BioBERT is presented. Proposed methods achieve best recall and F2 scores negating the need for any domain specific predefined linguistic patterns. This is a novel effort towards automatically generating interpretable patterns of connectivity for extracting 'connected' brain region mentions. The work has immediate relevance in the field of neuroscience, in that, it creates a repository of automatically compiled connections which can be used by researchers to validate the findings of the wet lab experiments with community reporting. Implementation of the proposed algorithms is hosted as a web application for use by neuroscientists on a large repository of over 50,000 full-text PubMed articles.