Identifying Depressive Symptoms from Tweets: Figurative Language Enabled Multitask Learning Framework
Published in International Conference on Computational Linguistics (COLING), 2020
Recommended citation: Yadav, Shweta, Jainish Chauhan, Joy Prakash Sain, Krishnaprasad Thirunarayan, Amit Sheth, and Jeremiah Schumm. "Identifying Depressive Symptoms from Tweets: Figurative Language Enabled Multitask Learning Framework." In Proceedings of the 28th International Conference on Computational Linguistics, pp. 696-709. 2020. https://aclanthology.org/2020.coling-main.61.pdf
Existing studies on using social media for deriving the mental health status of users focus on the depression detection task. However, for case management and referral to psychiatrists, health-care workers require a practical and scalable depressive disorder screening and triage system. This study aims to design and evaluate a decision support system (DSS) to reliably determine the depressive triage level by capturing fine-grained depressive symptoms expressed in user tweets through the emulation of the Patient Health Questionnaire-9 (PHQ-9) that is routinely used in clinical practice. As the 280-character limit on tweets incentivizes the use of creative artifacts in the utterances and figurative language forms a general fabric of communication for effective expression, the reliable detection of depressive symptoms from tweets is challenging. We propose a novel BERT based robust multi-task learning framework to accurately identify the depressive symptoms using the auxiliary task of figurative language detection. Specifically, our proposed novel task-sharing mechanism, co-task aware attention, enables automatic selection of optimal information across the BERT layers by soft-sharing of parameters. Our results show that modeling figurative language can demonstrably improve the model’s robustness and reliability for distinguishing the depression symptoms. [Download paper here] (https://aclanthology.org/2020.coling-main.61.pdf)