Depression is a common mental illness that can greatly impact an individual's ability to function when the condition is severe. With the proliferation of social media platforms, more and more studies have started to analyze users' social media data to detect their depression. However, most of the existing studies ignore information about users' cognitive distortions. In addition, these studies focus only on the semantic information of users' posts and ignore a variety of user-level information related to depression. To address these weaknesses, we propose a depression detection model fusing cognitive distortions and user-level information. On the one hand, the model captures potential multi-granularity cognitive distortion in formation in user text via a cognitive distortion encoder trained by a multi-granularity cognitive distortion learning method. In addition, we propose a post-level distortion aware mechanism to identify key cognitive distortion in formation related to the subjective cognition of users. On the other hand, the model fuses user-level information that is highly correlated with depression symptoms. It also dynamically identifies key post-level fusion information and multiple user-level information for different users through a multi-feature adaptive weighting method, which reduces noise from data irregularities. The results of the experiments show that the model achieves state-of-the-art performance on a public dataset. In addition, we discuss the impact of different cognitive distortion learning strategies and the importance of different user-level information in detecting depression through various experiments.
Wan et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: