Twitter-Based Early Depression Detection Through Machine Learning

Depression, a prevalent mental health disorder, significantly affects individuals’ quality of life, emphasizing the critical need for early detection and intervention to ensure successful treatment. However, conventional methods for early depression detection have limitations that can impact accuracy and effectiveness. The proposed research advocates integrating technology to overcome these challenges, explicitly leveraging social media and machine learning. The focus is on detecting early signs of depression in Twitter users. A dataset of tweet texts is gathered and employed in machine learning classifiers, including Naïve Bayes, Logistic Regression, and Support Vector Machine (SVM). The research aims to identify the most effective algorithm for depression detection by comparing testing results and determining the classifier with the highest accuracy level. This contributes to achieving Sustainable Development Goal 3 (SDG3) by enhancing individuals’ quality of life. The result proved that the Support Vector Machine (Linear Support Vector Classifier) is best performed with the highest accuracy value of 90.57% among the three tested classification models.