Connectome-based predictive models using resting-state fMRI for studying brain aging.

Area of Science:

• Neuroscience
• Radiology
• Gerontology

Background:

• Brain structure and function change with age, offering potential biomarkers for chronological age.
• Resting-state functional magnetic resonance imaging (fMRI) can capture functional brain connectomes.
• Functional connectomes have shown promise as predictors of individual age.

Purpose of the Study:

• To investigate the utility of connectome-based predictive modeling (CPM) using resting-state fMRI for predicting chronological age.
• To identify specific brain regions and networks that are significant predictors of age.
• To validate the generalizability of the developed age prediction model on independent datasets.

Main Methods:

• Applied connectome-based predictive modeling (CPM) to resting-state fMRI data from open-source datasets.
• Utilized the Southwest University Adult Lifespan Dataset for model training and feature identification.
• Validated the model's generalizability using independent datasets from the Autism Brain Imaging Data Exchange I and Open Access Series of Imaging Studies 3.

Main Results:

• CPM successfully predicted chronological age using resting-state fMRI data.
• Significant predictive features included a positive connection from the left inferior parietal sulcus and a negative connection from the right middle temporal sulcus.
• The subcortical-cerebellum network emerged as the dominant network for age prediction.

Conclusions:

• Connectome-based predictive modeling (CPM) using resting-state fMRI is a robust method for predicting chronological age.
• Specific functional brain connections and the subcortical-cerebellum network are key indicators of brain aging.
• The CPM approach demonstrates generalizability across different datasets, highlighting its potential clinical applicability.