Using modular connectome-based predictive modeling to reveal brain-behavior relationships of individual differences...

Area of Science:

• Neuroscience
• Cognitive Neuroscience
• Brain Imaging

Background:

• Working memory is essential for daily cognitive functions.
• Brain imaging techniques, particularly functional magnetic resonance imaging (fMRI), are increasingly used to predict working memory capacity.
• Existing predictive models have limitations in interpretability and generalization.

Purpose of the Study:

• To develop an improved connectome-based predictive modeling (CPM) approach for predicting individual working memory performance.
• To enhance the interpretability and accuracy of predictive models derived from whole-brain functional connectivity.
• To validate the model's performance on diverse cognitive tasks and external datasets.

Main Methods:

• Utilized n-back task-based and resting-state fMRI data from the Human Connectome Project (HCP).
• Employed an improved connectome-based predictive modeling (CPM) strategy to build predictive models.
• Performed anatomical feature analysis and compared prediction effects across different brain networks.

Main Results:

• The developed CPM approach demonstrated superior interpretability compared to prior models.
• The model showed a stronger alignment with known anatomical and functional brain networks.
• The model exhibited strong generalization capabilities across nine other cognitive behaviors within the HCP database and predicted working memory performance in external datasets.
• Identified specific brain networks crucial for distinguishing between high and low working memory load conditions.

Conclusions:

• The improved CPM approach provides a more interpretable and accurate method for predicting working memory performance from brain connectivity.
• The findings highlight the importance of specific brain networks in modulating working memory load.
• This approach has potential applications in understanding and potentially diagnosing cognitive impairments.