A focus of our lab is the identification of individuals that are at-risk to develop a psychiatric disorder. This approach offers the opportunity to prevent disorders before they occur in full effect and also to provide those individuals with treatment that need it most. The clinical-high risk state is an established syndrome that is associated with increases risk to develop psychosis. Recent findings indicate that machine learning tools can be used to predict outcome (Koutsouleris et al., 2018) and transition to psychosis in these individuals. However, before such tools can be translated to clinical practice, many questions need to be addressed. Currently, we are trying to answer the question of what data modalities are most promising to provide robust and reliable predictions in those individuals and how robust those models are if translated to different clinical contexts.