Abstract
Understanding sex differences in the adolescent brain is crucial, as they relate to sex-specific neurological and psychiatric conditions. Predicting sex from adolescent brain data may reveal how these differences influence neurodevelopment. Recently, attention has shifted toward socially-identified gender (distinct from sex assigned at birth) recognizing its explanatory power. This study evaluates whether resting-state functional connectivity (rsFC), cortical thickness, or cortical volume better predicts sex and sex/gender alignment (congruence between sex and gender) in preadolescents. Using Adolescent Brain Cognitive Development (ABCD) Study data and machine learning, rsFC predicted sex more accurately (85 %) than cortical thickness (76 %) and cortical volume (70 %). Brain regions most predictive of sex belonged to association (default mode, dorsal attention, parietal memory) and visual networks. The rsFC classifier trained on sex/gender aligned youth classified more accurately unseen youth with sex/gender alignment (n = 2013) than unalignment (n = 1116). The female rsFC sex profile was positively associated with sex/gender alignment, while in males, there was a negative association. However, neither brain modality predicted sex/gender alignment. These findings suggest that while rsFC predicts sex in the adolescent brain more accurately, it does not directly capture sex/gender alignment, underscoring the need for further investigation into the neural underpinnings of gender.