From Detection to Empowerment: Integrating a context-aware coping strategies recommendations tool into an automatic depression detection system in social networks

The aim of this study is to propose an automated system for the early identification of non-clinical depression-related behaviors among social media users. We constructed a novel dataset comprising 22,525 posts labeled by expert annotators. These posts are divided into 10,573 depressive and 11,952 non-depressive entries. Within the depressive category, we delineated two levels: moderate and major depression.

Subsequently, to identify users experiencing depression based on their posted content, we integrated various feature extraction techniques with machine learning and deep learning algorithms, including Support Vector Machine, K-Nearest Neighbors, Decision Tree, Recurrent Neural Network, and Long Short-Term Memory. After detecting depressed users, our system offers tailored recommendations for activities and coping strategies based on factors such as the severity of the case, gender, and age. These personalized recommendations can significantly contribute to the users’ healing journey, overall well-being, and especially in fostering a sense of hope and support. Experimental results indicate that our proposed system achieved an 84% accuracy rate in depression detection.