Artificial intelligence (AI) is opening a new frontier for proactive, evidence-based mental health support—particularly in digital ecosystems where early signals of distress, self-harm risk, harassment, or social exclusion may emerge long before clinical contact. This research theme focuses on designing robust, ethical, and clinically meaningful AI methods to detect and model mental-health risk patterns from real-world multimodal behavioural data, with special attention to adolescents and online social environments.
The project will develop multimodal and temporal predictive models that integrate heterogeneous signals (e.g., language, interaction patterns, social network structure, digital activity rhythms, and contextual metadata). A central objective is to bridge computational modelling with psychological and behavioural theories, enabling models that are not only accurate, but also scientifically interpretable and actionable for practitioners.
Strong emphasis will be placed on responsible AI: privacy-preserving learning, bias and fairness auditing, transparent explanations, uncertainty estimation, and governance aligned with healthcare standards and European regulatory expectations.
Keywords
Multimodal Mental Health AI; Digital Behaviour Modelling; Social Media Analytics; Harassment/Bullying Detection; Human-in-the-Loop AI; Explainable and Fair ML; Ethical AI and Clinical Governance.
Principal Investigator
Jordi González leads projects on smart mental health solutions through multimodal data analysis.
Email: jordi.gonzalez@cvc.uab.cat
Web: https://www.cvc.uab.es/research-lines/ise/
