Connectome-based predictive modeling shows sex differences in brain-based predictors of memory performance

Area of Science:

• Neuroscience
• Cognitive Aging
• Neuroimaging

Background:

• Alzheimer's disease (AD) disproportionately affects women, exhibiting higher prevalence and faster progression.
• Amnestic AD targets the default mode network (DMN), crucial for short-term memory.
• Previous studies indicate greater posterior DMN connectivity in aging women, suggesting increased reliance on this network.

Purpose of the Study:

• To investigate sex-based differences in brain network predictors of memory performance in aging.
• To specifically examine the role of the default mode network (DMN) in sex-specific memory prediction.

Main Methods:

• Applied connectome-based predictive modeling (CPM), a machine learning technique, to the Lifespan Human Connectome Project-Aging (HCP-A) dataset (n=579).
• Evaluated predictive models using cross-validation across the entire group and separately for each sex.
• Focused on an associative memory task-based scan for optimal model performance.

Main Results:

• Whole-group models predicted short-term memory performance with accuracies ranging from ρ = 0.21–0.45.
• Sex-specific models revealed distinct connectome-based predictors for men and women.
• Women's memory scores were more strongly predicted by DMN activity, whereas men's scores were predicted by visual network activity. Women showed greater DMN segregation, while men exhibited more visual network segregation.

Conclusions:

• Women and men recruit different neural circuitry during memory tasks.
• Women's increased reliance on the DMN for memory recall may contribute to their higher risk and faster progression of Alzheimer's disease.
• These findings highlight sex-specific neural mechanisms underlying memory function and AD risk.