An Automatic Screening for Major Depressive Disorder from Social Media in Thailand

Abstract

The MDD patients do not often realize their own risk of being MDD and hardly access to treatments in its early state. Thus, widely screening test is necessary to prevent and control the mental disorder. This paper presents a method for screening major depressive disorder (MDD) from Thai posts in social media. Unlike using questionnaire which requires a person to personally perform, the proposed method can help in an MDD screening task to cover in mass-screening on social media. The machine-learning based classification is exploited to develop a model in discriminate positive and negative risk of being MDD according to Thai screening questionnaire(2Q). From evaluation, the best machine-learning technique for the task was support vector machine for 0.94 F-measure score. From the screening model, we also found top-10 Thai terms that are positively informative for a risk of being MDD.