Explainable Depression Detection with Multi-Modalities Using a Hybrid Deep Learning Model on Social Media
This work addresses the need for interpretable depression detection models for mental health applications on social media platforms, though it appears to be an incremental improvement combining existing techniques.
The paper tackles the problem of detecting depression from social media posts by proposing MDHAN, a hybrid deep learning model that uses multi-modal features and hierarchical attention mechanisms to provide interpretable predictions. The model outperforms several baseline methods and demonstrates improved predictive performance for depression detection on public social media messages.
Model interpretability has become important to engenders appropriate user trust by providing the insight into the model prediction. However, most of the existing machine learning methods provide no interpretability for depression prediction, hence their predictions are obscure to human. In this work, we propose interpretive Multi-Modal Depression Detection with Hierarchical Attention Network MDHAN, for detection depressed users on social media and explain the model prediction. We have considered user posts along with Twitter-based multi-modal features, specifically, we encode user posts using two levels of attention mechanisms applied at the tweet-level and word-level, calculate each tweet and words' importance, and capture semantic sequence features from the user timelines (posts). Our experiments show that MDHAN outperforms several popular and robust baseline methods, demonstrating the effectiveness of combining deep learning with multi-modal features. We also show that our model helps improve predictive performance when detecting depression in users who are posting messages publicly on social media. MDHAN achieves excellent performance and ensures adequate evidence to explain the prediction.