Predicting brain-age from multimodal imaging data captures cognitive impairment.

Area of Science:

• Neuroscience
• Biomarkers
• Aging

Background:

• Brain age prediction uses biological data to estimate brain maturity.
• Previous studies relied on single data types (structural or functional imaging).
• A multimodal approach may enhance brain age prediction accuracy.

Purpose of the Study:

• To investigate if multimodal brain imaging data improves brain age prediction.
• To assess if the brain-age gap correlates with cognitive impairment.
• To evaluate the robustness and generalizability of multimodal brain-age models.

Main Methods:

• Utilized cortical anatomy and functional connectivity data from a large adult sample (N=2354, ages 19-82).
• Developed multimodal brain-age prediction models.
• Tested model robustness against head motion and generalizability on an independent dataset (N=475).

Main Results:

• Multimodal data significantly improved brain age prediction accuracy (mean absolute error of 4.29 years).
• The brain-age gap effectively captured cognitive impairment.
• Models demonstrated robustness to head motion and reasonable generalization to external data.

Conclusions:

• Multimodal brain-age prediction is a robust and generalizable method.
• This approach shows potential for early prediction of neurocognitive disorders.
• Larger, heterogeneous datasets enhance model generalizability.