Predicting future cognitive decline from non-brain and multimodal brain imaging data in healthy and pathological...

Area of Science:

• Neuroscience
• Gerontology
• Medical Imaging

Background:

• Previous research focused on predicting diagnostic labels from brain imaging.
• Subtle brain changes precede cognitive decline in aging.
• This study predicts cognitive decline as a continuous trajectory.

Purpose of the Study:

• To determine if multimodal neuroimaging data improves prediction of future cognitive decline.
• To assess prediction accuracy in both healthy and pathological aging.

Main Methods:

• Utilized baseline data (demographics, clinical, neuropsychological scores, structural MRI, functional connectivity) from the OASIS-3 dataset (N=662).
• Employed cross-validated multitarget random forest models.
• Predicted future cognitive decline (CDR, MMSE) approximately 5.8 years ahead.

Main Results:

• Combining non-brain data with structural MRI significantly improved continuous prediction of cognitive decline (R²=0.42).
• Cognitive performance, daily functioning, and subcortical volume were key predictors.
• Functional connectivity data did not enhance predictive accuracy.

Conclusions:

• Baseline non-brain and structural MRI data can effectively predict future cognitive decline trajectories.
• Prognosis of age-related cognitive decline can inform personalized interventions.
• Future research may enable earlier and more effective interventions for cognitive aging.