Explainable Multimodal-based Depression Awareness at the Edge

Depression is among the most prevalent and debilitating mental health conditions worldwide, with more than 230 million people suffering from major depressive disorder globally, making it a major contributor to the overall burden of disease. While automated diagnostic approaches have already been investigated to support this depressive burden, current research focuses mainly on black-box models that provide post-hoc explainability at best. Existing research into white-box approaches often considers only single modalities or does not account for the role of context at the time of data collection. Today’s state of depression monitoring thus still lacks explainable approaches that also consider multimodal and context-aware dimensions for household environments.