Detection of depressive comments on social media using RNN, LSTM, and random forest: comparison and optimization

Depression is a prevalent mental health condition, and social media platforms have become valuable sources for understanding individuals’ emotional well-being. In this study, we aimed to develop an accurate classification model for detecting depressive comments in social media data using different methods. Three different models were evaluated: Recurrent neural network (RNN) with Long short-term memory (LSTM), RNN with bidirectional LSTM, and Random Forest. Additionally, the performance was evaluated in terms of accuracy, precision, recall, F1-score, specificity, and sensitivity. The results showed that the optimized model achieved a precision of 83.32%; furthermore, the accuracy of the tuned model improved from 75.41 to 80.07%, demonstrating its potential for real-world applications in monitoring mental health trends and providing support to individuals experiencing depression.