Karcher, N. R. | Search bioRxiv

bioRxiv2025

Longitudinal Trajectories of Cognition and Neural Metrics as Predictors of Persistent Distressing Psychotic-Like Experiences Across Middle Childhood and Early Adolescence

ObjectivesPsychotic-like experiences (PLEs) may arise from genetic and environmental risk leading to worsening cognitive and neural metrics over time, which in turn lead to worsening PLEs. Persistence and distress are factors that distinguish more clinically significant PLEs. Analyses used three waves of unique longitudinal Adolescent Brain Cognitive Development Study data (ages 9-13) to test whether changes in cognition and structural neural metrics attenuate associations between genetic and environmental risk with persistent distressing PLEs. MethodsMultigroup univariate latent growth models examined three waves of cognitive metrics and global structural neural metrics separately for three PLE groups: persistent distressing PLEs (n=356), transient distressing PLEs (n=408), and low-level PLEs (n=7901). Models then examined whether changes in cognitive and structural neural metrics over time attenuated associations between genetic liability (i.e., schizophrenia polygenic risk scores/family history) or environmental risk scores (e.g., poverty) and PLE groups. ResultsPersistent distressing PLEs showed greater decreases (i.e., more negative slopes) of cognition and neural metrics over time compared to those in low-level PLE groups. Associations between environmental risk and persistent distressing PLEs were attenuated when accounting for lowered scores over time on cognitive (e.g., picture vocabulary) and to a lesser extent neural (e.g., cortical thickness, volume) metrics. ConclusionsAnalyses provide novel evidence for extant theories that worsening cognition and global structural metrics may partially account for associations between environmental risk with persistent distressing PLEs.

neuroscience↗

bioRxiv2023

Network level enrichment provides a framework for biological interpretation of machine learning results

Machine learning algorithms are increasingly used to identify brain connectivity biomarkers linked to behavior and clinical outcomes. However, non-standard methodological choices in neuroimaging datasets, especially those with families or twins, have prevented robust machine learning applications. Additionally, prioritizing prediction accuracy over biological interpretability has made it challenging to understand the biological processes behind psychopathology. In this study, we employed a linear support vector regression model to study the relationship between resting-state functional connectivity networks and chronological age using data from the Human Connectome Project. We examined the effect of shared variance from twins and siblings by using cross-validation, either randomly assigning or keeping family members together. We also compared models with and without a Pearson feature filter and utilized a network enrichment approach to identify predictive brain networks. Results indicated that not accounting for shared family variance inflated prediction performance, and the Pearson filter reduced accuracy and reliability. Enhancing biological interpretability was achieved by inverting the machine learning model and applying network-level enrichment on the connectome, while directly using regression coefficients as feature weights led to misleading interpretations. Our findings offer crucial insights for applying machine learning to neuroimaging data, emphasizing the value of network enrichment for comprehensible biological interpretation.

neuroscience↗