Combining longitudinal change features of childhood psychopathology with Polygenic scores in machine learning models of adult wellbeing

Combining multiple modalities of data for the prediction of wellbeing in machine learning models might lead to a more accurate prediction. This project aims to investigate the usefulness of longitudinal features of a specific aspect of an individual’s life history, namely childhood psychopathology, for machine learning-based prediction of adult wellbeing. Features derived from longitudinal trajectories of childhood psychopathology will be compared to polygenic risk scores for a variety of phenotypes and to cross-sectional features of childhood psychopathology ignoring the longitudinal aspect. It is expected that longitudinal features will be of high feature importance.