AI Models for Depressive Disorder Detection and Diagnosis: A Review
An extensive review encompassing 55 significant studies on artificial intelligence techniques for identifying and diagnosing depression has been conducted. This paper establishes a structured taxonomy categorizing the domain based on clinical tasks (diagnosis versus prediction), types of data (text, speech, neuroimaging, and multimodal), and classes of computational models (such as graph neural networks, large language models, and hybrid methods). Notable trends include the dominance of graph neural networks in modeling brain connectivity, the increasing use of large language models for text and conversational data, and a growing emphasis on multimodal integration. The analysis underscores AI's potential to create objective, scalable, and timely diagnostic tools for Major Depressive Disorder, a leading global cause of disability that currently depends on subjective clinical evaluations.
Key facts
- Major Depressive Disorder is a leading cause of disability worldwide.
- Current diagnosis depends largely on subjective clinical assessments.
- The survey reviews 55 key studies on AI methods for depression detection.
- A novel hierarchical taxonomy is introduced based on clinical task, data modality, and model class.
- Graph neural networks are predominant for modeling brain connectivity.
- Large language models are rising for linguistic and conversational data.
- An emerging focus on multimodal integration is noted.
- AI holds promise for developing objective, scalable, and timely diagnostic tools.
Entities
Institutions
- arXiv