Beyond discrete classifications: a computational approach to the continuum of cognition and behavior in children.

Abstract

Psychiatry is undergoing a shift toward precision medicine, demanding personalized approaches that capture the complexity of cognition and behavior. Here, we introduce a novel referential of four robust, replicable, and generalizable cognitive and behavioral profiles. These were derived from a large pediatric cohort (ABCD: n = 10,843) and validated in two independent cohorts (BANDA: n = 195 and GESTE: n = 271) regrouping children aged 9-17 years. We demonstrate the profiles' longitudinal stability and consistency with clinical diagnoses in the general population while exposing critical discrepancies across parent-reported, youth-reported, and expert-derived diagnoses. Beyond validation, we showcase the real-world utility of our approach by linking profiles to environmental factors, revealing associations between parental influences and youths' cognition and behavior. Our fuzzy profiling framework moves beyond discrete classification, offering a powerful tool to refine psychiatric evaluation and intervention. We provide an open-source framework, enabling researchers and clinicians to fast-track implementation and foster a data-driven, domain-based approach to diagnosis.