A new study published in PNAS used advanced AI to examine whether male and female brains can be differentiated by analyzing brain activity patterns. Srikanth Ryali and colleagues found significant and consistent differences between the sexes in certain brain areas, suggesting that these differences are meaningful and have implications for personalized medical approaches.
Sex plays an important role in brain development, aging, and the onset and progression of psychiatric and neurological disorders. For example, females are more likely to suffer from depression, anxiety, and eating disorders, while conditions such as autism, ADHD, and schizophrenia are more common in males, often with sex-specific symptoms and outcomes.
This study utilized spatiotemporal deep neural networks (stDNN) to distinguish sex differences within the Human Connectome Project (HCP) cohort by analyzing fMRI time series data. These models achieved high accuracy rates (90.21 to 91.17%) and demonstrated strong performance across various metrics including macro-precision, macro-recall, macro-F1 scores, and AUC. These results were consistent across different sessions, indicating reliable sex differences in brain function without the need for additional model training.
The distinctiveness and generalizability of brain features underlying sex differences were further explored through explainable AI (XAI) techniques and consensus analysis. Individual “fingerprints” identified through the IG procedure revealed specific brain regions—such as the precuneus and ventromedial prefrontal cortex—consistently distinguished between sexes.
These findings were validated across multiple datasets and independent cohorts. The study extended its analysis to cognitive functions, identifying sex-specific neurobiological predictors of cognition through canonical correlation analysis. Certain brain regions differentially predicted cognitive profiles in females and males.
This study introduces a groundbreaking approach using stDNN to identify sex differences in brain function, utilizing resting-state fMRI data. By learning directly from raw data, this method bypasses traditional reliance on predefined brain connectivity features. The research confirms intrinsic organizational differences between sexes and links these differences to cognitive functions, challenging prior understandings and suggesting sex-specific brain organization influences behavior.
The use of XAI revealed that specific brain areas, particularly the default mode network, striatum, and limbic system, play crucial roles in sex differences. These regions are associated with processes like self-referential thought and reward sensitivity, indicating that the functional disparities between sexes may underlie unique cognitive and behavioral patterns.
This link between brain organization and cognitive outcomes opens new avenues for sex-specific research in cognitive neuroscience and clinical work, emphasizing the importance of sex in the study of brain health and disorders.
The study, “Deep learning models reveal replicable, generalizable, and behaviorally relevant sex differences in human functional brain organization”, was authored by Srikanth Ryali, Yuan Zhang, Carlo de los Angeles, Kaustubh Supekar, and Vinod Menon.