DisorBERT: A Double Domain Adaptation Model for Detecting Signs of Mental Disorders in Social Media

Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics Volume 1: Long Papers, pages 15305–15318 July 9-14, 2023 ©2023 Association for Computational Linguistics DisorBERT: A Double Domain Adaptation Model for Detecting Signs of Mental Disorders in Social Media Mario Ezra Aragónα β, A. Pastor López-Monroyγ, Luis C. Gonzálezδ David E. Losadaα, Manuel Montes-y-Gómezβ α Centro Singular de Investigación en Tecnoloxias Intelixentes (CiTIUS), Universidade de Santiago de Compostela, Spain β Instituto Nacional de Astrofísica, Óptica y Electrónica (INAOE), Mexico γ Centro de Investigación en Matemáticas (CIMAT), Mexico δ Facultad de Ingeniería, UACh, Mexico {ezra.aragon,david.losada}@usc.es, pastor.lopez@cimat.mx, lcgonzalez@uach.mx, mmontesg@inaoep.mx Abstract Mental disorders affect millions of people worldwide and cause interference with their thinking and behavior. Through the past years, awareness created by health campaigns and other sources motivated the study of these dis- orders using information extracted from social media platforms. In this work, we aim to con- tribute to the study of these disorders and to the understanding of how mental problems reflect on social media. To achieve this goal, we pro- pose a double-domain adaptation of a language model. First, we adapted the model to social media language, and then, we adapted it to the mental health domain. In both steps, we incor- porated a lexical resource to guide the masking process of the language model and, therefore, to help it in paying more attention to words related to mental disorders. We have evaluated our model in the detection of signs of three ma- jor mental disorders: Anorexia, Self-harm, and Depression. Results are encouraging as they show that the proposed adaptation enhances the classification performance and yields competi- tive results against state-of-the-art methods. 1 Introduction Mental disorders are among the most common ill- nesses worldwide. Some estimates1 indicate that more than 50% of the population will be diagnosed with a mental disorder at some point in their lives. The prevalence of these disorders is highly con- cerning since they alter the way people think, feel, and take action, resulting in the incapacity of daily life routines. In addition, the recent COVID-19 pandemic triggered a serious global social and eco- nomic disruption, which had a direct effect on peo- ple’s lives, and brought many challenges that can be 1Center for Disease Control and Prevention, https://www.cdc.gov/mentalhealth/learn/index.htm stressful and overwhelming (Li et al., 2020). This situation was particularly difficult for people with mental health conditions and caused an increase in the prevalence of anxiety and depression (World Health Organization, 2022). There is therefore an increasing need for developing new tools to moni- tor the presence of mental disorders and to respond to early signs of psychological concerns. Nowadays, social media content is massive and provides an opportunity to do research on how peo- ple undergo difficulties. Many people use online platforms to publicly share their daily routines and important events, while others take advantage of the anonymity of these spaces to explicitly discuss mental health issues and to seek help (Ríssola et al., 2021; Crestani et al., 2022). In this work, …

Content truncated to 500 words