Mental health disorders, including depression, anxiety, and related conditions, are recognised as major global health challenges that affect quality of life and place a substantial burden on society. Current diagnostic practice depends heavily on clinician-led interviews and standardised questionnaires, making assessments time-consuming, subjective, and vulnerable to regional shortages of specialised professionals. Subtle affective signals, such as vocal tone, facial expression, and physiological responses, often remain unquantified, leading to under-diagnosis or misdiagnosis and delaying timely intervention. There is therefore a pressing need for objective, data-driven tools that can broaden access to accurate and early screening.

Recent developments in artificial intelligence offer powerful means to meet this need. Affective computing and multimodal learning enable the integration of diverse data streams, ranging from speech and text to video and bio-signals collected in daily life or clinical settings. Multimodal foundation models and large language models provide a unified framework for analysing these heterogeneous inputs, while knowledge-based techniques, including psychiatric ontologies, symptom networks, and retrieval-augmented generation, anchor predictions in established clinical evidence. By combining rich behavioural data with structured medical knowledge, AI systems can deliver diagnostic support that is not only sensitive and robust but also transparent and trustworthy. Furthermore, such systems can be adapted to different demographic groups, improving fairness and inclusiveness in mental health care.

This special issue aims to highlight research that advances multimodal, knowledge-driven, and explainable AI for mental-health diagnostics. We welcome original studies, surveys, and resource papers addressing (but not limited to) the following topics:

• Affective computing for mental health prediction
• Deep neural networks for affective state analysis
• Large Language Models (LLMs) in clinical mental health applications
• Multimodal foundation models for psychological diagnosis
• Retrieval-Augmented Generation (RAG) for explainable clinical support
• Knowledge engineering for psychiatric diagnosis
• Knowledge-based explainable AI systems
• Early detection systems for anxiety, depression, and related disorders