Detecting and Characterizing Mental Health Using Social Media Analytics

This paper aims to characterize and detect mental health from Twitter users’ posts and how the language, user attributes, and tweet attributes are associated with one of the most prevalent mental illnesses, depression. In this study, the social media analytics CUP framework and the Twitter API are used for the data collection of 27408 unique users. Data analysis uses expert input, natural language processing, and statistical methods. The empirical result from the statistical test confirms that linguistic features can represent the social media user’s mental illness conditions and attributes of social media posts, and users are significantly associated with mental illness and depression. Mental health is a widespread issue worsened by the pandemic. There’s still a stigma attached to it, which discourages open discussions. Digital technology, including social media, can significantly impact our mental health. Hence, the passive mechanism of mental illness characterization and detection from social media can help in timely intervention and provide the necessary support.