Background
Early identification of young people at risk of developing severe mental disorders remains a major challenge. Although machine learning and AI have advanced rapidly, few prediction models have translated into clinical use due to limited data availability, poor generalisability, lack of interpretability and insufficient involvement of people with lived experience. Integrating diverse multimodal data sources such as clinical, cognitive and functioning measures, imaging, speech, activity and other biomarkers offers new opportunities to improve prediction, enable earlier intervention and enhance understanding of underlying mechanisms.
Novelty and Importance
This project will combine modern statistical, machine learning and AI approaches with applied clinical research to develop and evaluate robust prediction models that dynamically estimate functioning and the onset of psychosis in high-risk individuals and related outcomes. It will draw on a large, well-curated longitudinal multimodal dataset from the Accelerating Medicines Partnership Schizophrenia programme. A distinctive feature of the project is the integration of methodological rigour with patient and public involvement, ensuring that model outputs are both technically sound and clinically meaningful while assessing fairness and bias. The work aligns with current priorities in precision medicine and responsible AI in mental healthcare.
Aims and Objectives
The project aims to (1) develop and validate dynamic multimodal prediction models using traditional regression and modern machine learning or AI methods; (2) compare model performance, calibration and interpretability across approaches; (3) explore causal and fairness-aware extensions to enhance clinical relevance; and (4) engage people with lived experience to guide model framing, interpretation and dissemination. The student will gain expertise in statistical modelling, AI and applied health data science, supported by secure cloud computing resources, multidisciplinary supervision and participation in specialist seminars and training activities.