A Connectivity-based Psychometric Prediction Framework for Brain-behavior Relationship Studies

The recent availability of population-based studies with standard neuroimaging measurements and extensive psychometric characterization opens promising perspectives to investigate the relationships between interindividual variability in brain regions' connectivity and behavioral phenotypes. However, the multivariate nature of the prediction model based on connectivity within a network of brain regions severely limits the interpretation of the brain-behavior patterns from a cognitive neuroscience perspective. To address this issue, we here propose a connectivity-based psychometric prediction (CBPP) framework based on individual region's connectivity profile. Preliminary to the development of this region-wise machine learning approach, we performed an extensive assessment of the general CBPP framework based on whole-brain connectivity information. Because a systematic evaluation of different parameters was lacking from previous literature, we evaluated several approaches pertaining to the different steps of a CBPP study. We hence tested 72 different approach combinations in a cohort of over 900 healthy adults across 98 psychometric variables. Overall, our extensive evaluation combined to an innovative region-wise machine learning approach, offering a framework that optimizes both prediction performance and neurobiological validity (and hence interpretability) to study brain-behavior relationships.