RDoCer: Information Retrieval and Sentence Extraction for Mental Health using Research Domain Criteria

Research Domain Criteria (RDoC) is a classification framework for mental illness, recently introduced by the National Institute of Mental Health. The RDoC Task at this years' BioNLP Open Shared Tasks 2019 workshop is a competition for inviting text mining groups around the world for developing informatics models for two subtasks: information retrieval and sentence extraction for mental health using RDoC. For competing in the RDoC task, we developed RDoCer, which uses a Term Frequency Inverse Document Frequency-based similarity measure for information retrieval and supervised machine learning models for sentence extraction. In comparison to the other models competed at the RDoC task, RDoCer performs competitively across the two subtasks while notably ranking first for the Sustained Threat RDoC construct in the sentence extraction subtask. The findings of this study have implications for mental health informaticians as well as researchers, curators, clinicians working in this domain.