Development and application of machine learning and statistical methods to investigate the relation between exposome and mental health and cognitive development in children.

This PhD project is part of the Equal-Life project: an EU-funded, international and interdisciplinairy project on the subject on the effect of the exposome, from conception to young adulthood, on mental health and cognition outcomes, This PhD project’s first goal is the validation, optimization and application of Random Forest-based feature importance rankings to elucidate the relationship between the early-life exposome and outcomes in the domains of mental health and cognition. Once this part of the project is completed, the project will look into additional statistical issues and knowledge gaps of relevance to the Equal-Life project. Proposed subjects are the pooling of results from different studies when the included features do not fully overlap and the creation of a neighbourhood characterization algorithm. To this end, simulated datasets will be used to investigate the effects of different data characteristics on the quality of feature importance rankings based on different types of Random Forest methods. The newly obtained knowledge will then be applied to create and evaluate feature importance rankings for mental health and cognition outcomes of least two Dutch cohort studies and one Dutch cross-sectional study. Additionally, feature importance rankings from different studies will be compared to further validate the rankings and spot inconsistencies.