Aristotle University Of Thessaloniki (AUTH)
Predicting mental health indices with artificial intelligence and machine learning
Abstract
dc:descriptionThis dissertation explores the integration of behavioral features derived from the Relative Preference Theory (RPT) and explainable machine learning to advance the prediction of psychiatric conditions. Leveraging a novel application of the RPT, where interpretable metrics are derived from picture rating tasks that model human judgment and preference patterns. Four datasets, collected before and during the COVID-19 pandemic, provide the foundation for this research, enabling an analysis of the pandemic's impact on mental health and the efficacy of RPT features in predicting conditions such as anxiety, depression, suicidality, psychotic symptoms etc. A total of eleven studies demonstrate the feasibility and utility of RPT features in ML models, emphasizing the importance of explainability in artificial intelligence applications. By employing techniques like feature selection, exploratory data analysis, and minimal feature set optimization, the research achieves a balance between computational efficiency and interpretability. This ensures that the developed predictive models are both robust and clinically actionable, addressing challenges in MH diagnostics, particularly in underserved populations. The findings underscore the transformative potential of computational psychiatry. The research reveals that the COVID-19 pandemic significantly exacerbated MH conditions, highlighting the critical role of early detection and intervention. By validating the utility of RPT metrics in diverse contexts, the dissertation establishes a framework for future studies aimed at fusing behavioral, biological, and environmental data to enhance predictive accuracy. This work contributes to the democratization of MH care by providing scalable, interpretable tools for identifying psychiatric disorders. It sets the stage for precision psychiatry, where AI-driven insights can guide tailored interventions, improving outcomes and addressing global disparities in mental health services.
Degree
thesis:*- Grantor dc:publisher
- Aristotle University Of Thessaloniki (AUTH)
- Year dc:date
- 2025
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Stefanopoulos, Leandros
- Στεφανόπουλος, Λέανδρος
Subjects
dc:subject × 13- Θεωρία σχετικών προτιμήσεων
- Επεξηγήσιμη μηχανική μάθηση
- Υπολογιστική ψυχιατρική
- Relative preference theory
- Explainable machine learning
- Computational psychiatry
- Computational cognition
- Επιστήμες Υγείας
- Ιατρική και Επιστήμες Υγείας
- Ιατρική Πληροφορική
- Health Sciences
- Medical and Health Sciences
- Medical Informatics
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/58199