An artificial intelligence based ensemble learning model for stress detection based on social media interactions

Depression is an important mental health problem that has increased with the widespread use of social media and negatively affects the quality of life of individuals. According to the World Health Organization, early diagnosis of depression plays a critical role in preventing serious consequences such as suicide. This thesis aims to analyze the symptoms of depression and anxiety on users' moods through posts shared on the social media platform. Within the scope of the study, a method based on sentiment analysis was developed by identifying the words and expressions frequently used by users with depressive tendencies. This study proposes a model for the effective use of social media data for early diagnosis and intervention of depression. The analysis of social media data focuses on the detection of depression symptoms and for this purpose, the basic concepts and methods of sentiment analysis are discussed in detail. In the study, machine learning algorithms such as support vector machines, artificial neural networks and XGBoost are applied on the data obtained from text mining processes. In addition, ensemble learning approaches were used to improve model performance and the accuracy rates of these methods were compared with similar studies in the literature. The results show that the proposed model provides higher accuracy and performance in depression detection compared to traditional methods. The performance metrics of the proposed model are accuracy 91%, precision 90%, recall 91.5% and F1-score 0.907. In this context, the study shows that making sense of social media data through sentiment analysis can offer innovative approaches in the field of mental health and provide a valuable data source for healthcare professionals.