Interpretable brain age prediction using linear latent variable models of functional connectivity.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Data Science

Background:

• Brain age prediction from neuroimaging is a growing research area.
• Understanding age-related changes in functional connectivity is crucial.
• Existing models often lack interpretability.

Purpose of the Study:

• To develop interpretable models for brain age prediction using functional connectivity.
• To investigate how functional connectivity changes with age.
• To ensure generalizability of the developed models.

Main Methods:

• A two-step approach was used: first, linear latent variable models (e.g., PCA) identified reproducible functional connectivity networks.
• Second, network activity served as features in a linear regression model to predict brain age.
• The framework was applied to CamCAN, HCP, and ATR datasets.

Main Results:

• The models accurately predicted brain age.
• The approach allowed for the investigation of age-related functional connectivity patterns.
• The models demonstrated generalization across different neuroimaging datasets.

Conclusions:

• Interpretable brain age prediction models based on functional connectivity are feasible.
• This framework aids in understanding neurobiological aging processes.
• The method shows promise for clinical and research applications in aging.