(1) Neurological and neurodegenerative diseases are collectively a major public health burden with sex-based disparities.

(2) Predictive and diagnostic models of the aging brain that integrate blood- and brain markers will advance our understanding of the aging brain in health and disease

(3) The high-dimensionality of multi-parametric MRI can reduce analytical sensitivity and highlight the need for latent biomarkers of brain health, such as brain age gap.

• Our Brain Age model outperforms state-of-the-art approaches, including image-based models, achieving a MAE of 2.50 while remaining sensitive to different neurodegenerative disorders across diagnostically relevant feature sets.

• Models that are inclusive of WM microstructural features, derived from diffusion MRI, are most sensitive in the detection of demyelinating diseases and traumatic brain injury.

• We demonstrate the association of BAG model predictions with plasma proteins, while also identifying 20 novel proteins that may have a more sex-specific effect on brain aging.

• Alignment between SHAP feature importance and the underlying pathology of ADRD and demyelinating diseases suggests that the models are capturing biologically relevant signals

• Together, this work highlights sex-specific blood-based targets to inform early, stratified brain health interventions.