Detecting Depression in Arabic Texts Using AI Techniques

In Arabic-speaking communities, mental health issues often go unrecognized due to cultural stigma and a lack of awareness, making early diagnosis challenging. Social media platforms and online forums provide a rich source of data that can be used to identify early signs of depression through analysis of user-generated content. This study aimed to develop a reliable method for detecting depressive tendencies in Arabic texts using artificial intelligence techniques. To achieve this, we collected approximately 19,000 posts from the Arabic forum “Nafsany,” using Python's BeautifulSoup library to scrape data based on keywords related to depression. The data were annotated with the help of an expert psychologist and classified according to the DSM-5 guidelines. Our study employed lexicon-based, machine learning, and deep learning approaches. An ensemble model combining various deep learning techniques was developed to enhance the accuracy of depression detection. The ensemble model achieved superior performance, with an accuracy of 93.03% and an F1-score of 92.93%, which demonstrates its effectiveness. This research is significant as it contributes to the early detection of depression in Arabic-speaking populations, potentially reducing stigma and barriers to mental health care. By improving the diagnostic methods, this study supports better mental health outcomes and provides a foundation for future research in the field.