Depression classification on social media with deep learning

The detection of mental illnesses on social media can be considered a complex task primarily due to the intricate nature of mental disorders. In recent years, social media networks have become an integral part of our daily lives. Acquiring information, tracking trends, and understanding the emotions of individuals on social media have become significantly important. Correspondingly, the burgeoning popularity of social media platforms has led to the development of this research area. There exists a close relationship between social media platforms and their users, thereby reflecting various aspects of users' personal lives on these platforms. In such an environment, researchers are presented with rich information about human life. With advancements in machine learning and the availability of sample data related to depression, there lies the possibility of developing an early depression diagnosis system a key to reducing the number of individuals suffering from depression. This study proposes an efficient model by employing a deep learning model, Bidirectional Long Short-Term Memory (Bi-LSTM), to predict depression by scraping tweets using an API to create its own dataset. Natural language processing techniques were applied to prepare the data for machine learning training procedures. The proposed framework achieves higher accuracy, reaching 97.25%, compared to other deep learning and machine learning models, while also reducing the false positive rate. The suggested model is compared with other models in terms of average accuracy. This proposed approach demonstrates successful results in the early detection of depression in Twitter users' sentiments, showcasing the applicability of Bi-LSTM.