PRECOGNITION: Learning latent cognitive profiles to predict psychosis

Cognitive impairments are frequently evident early in the course of psychosis, often before illness onset and having substantial impact on functional outcomes. However, this has not yet produced markers with sufficient sensitivity and specificity to predict outcomes at the level of the individual or facilitate early intervention. This project will address this need by applying normative models ('cognitive growth charting') to cognitive measures derived from tens of thousands of healthy individuals and thousands of individuals with psychosis from four European countries in order to infer early risk factors and predict functioning in psychosis at the individual level. We will go beyond symptoms and assess functioning broadly using measures that meaningfully reflect the quality of life of individuals with psychosis. We will: (i) develop deep learning technology to enable cognitive data from studies with heterogeneous cognitive data to be aggregated; (ii) map lifespan variation across multiple cognitive domains and precisely place individuals within population norms (iii) integrate these with neurobiology in order to precisely stratify cohorts, predict progression at the individual level and identify genetic and environmental factors that modulate developmental trajectories both preceding and following illness onset. Our project will be guided throughout by extensive engagement with lived experience experts.