Prediction studies in early psychosis and related research has generally relied on one-off sampling of cross-sectional data (i.e., a "snapshot" of clinical state and other risk markers at study entry). A new approach is emerging based on the view that psychopathology is an evolving, dynamic picture, particularly in the early stages of disorder, and that predictive modelling may therefore be maximised by taking these temporal changes into account. This approach to prediction research in psychopathology draws on cross-disciplinary models of complex systems and advances in statistical modelling. This symposium consists of presentations illustrating this emerging paradigm in prediction research, consistent with the “broadening the scope” theme of the conference. Hok Pan Yuen will present joint modeling analysis of longitudinal clinical variables in an ultra high risk for psychosis cohort. Dr Wichers will present novel empirical support for the presence of early warning signals in anticipation of critical transitions in depressive symptomatology. Dr Hartmann will present data relating to early warning signals (psychopathology and circadian activity) from a cohort study of young people at risk of a range of mental disorders. Dr Wigman will present findings from a new diary study (daily recordings of psychopathology) mapping individual symptom networks and the predictive value of these for course/outcome. This next generation of prediction studies may more accurately identify people at highest risk of disorder progression or relapse, which clearly has important treatment implications, such as stratifying patients to particular treatments and introducing methods for identifying early warning signs of mental state deterioration.
