Related Experiment Video
Updated: Feb 19, 2026
05:55
Modeling the Functional Network for Spatial Navigation in the Human Brain
Published on: October 13, 2023
1.6K
Connectome-based predictive modeling of attention: Comparing different functional connectivity features and
Kwangsun Yoo1, Monica D Rosenberg1, Wei-Ting Hsu1
1Department of Psychology, Yale University, New Haven, CT, USA.
Neuroimage
|November 11, 2017
Summary
Connectome-based predictive modeling accurately predicts attention using brain connectivity data. Task-based functional connectivity measures and partial least square regression generally yielded the best predictions.
Area of Science:
- Neuroscience
- Cognitive Science
- Machine Learning in Neuroimaging
Background:
- Connectome-based predictive modeling (CPM) predicts individual differences from functional brain connectivity (FC) measured via fMRI.
- Previous studies have shown CPM's utility for predicting fluid intelligence and sustained attention.
- The current study investigates novel FC measures and prediction algorithms within the CPM framework for attention prediction.
Purpose of the Study:
- To compare the predictive power of different functional connectivity measures (Pearson's correlation, accordance, discordance) for attention.
- To evaluate the efficacy of different prediction algorithms (linear regression, partial least square [PLS] regression) for attention prediction.
- To determine the impact of using task-based versus resting-state fMRI data for building predictive models of attention.
Main Methods:
- Employed the Connectome-based predictive modeling (CPM) framework using fMRI data.
- Compared three functional connectivity (FC) measures: Pearson's correlation, accordance, and discordance.
- Utilized two prediction algorithms: linear regression and partial least square (PLS) regression, with internal and external cross-validation.
Main Results:
- All tested combinations of FC measures and algorithms successfully predicted attentional abilities, with high internal (r=0.9) and external (r=0.6) validation correlations.
- Models trained on task-based fMRI data demonstrated superior predictive performance compared to those trained on resting-state data.
- Pearson's correlation and accordance measures slightly outperformed discordance, while PLS regression generally yielded better results than linear regression.
Conclusions:
- CPM is a viable method for predicting individual differences in attention using brain connectivity.
- Task-based functional connectivity data, particularly using accordance measures with PLS regression, offers a promising approach for enhancing attention prediction.
- The findings support the consideration of accordance features and PLS regression for future CPM applications in attention research.
