Heterogeneity Analysis on Multi-state Brain Functional Connectivity and Adolescent Neurocognition

Area of Science:

• Neuroscience
• Developmental Neuroscience
• Computational Neuroscience

Background:

• Brain functional connectivity (connectome) is key to understanding behavior.
• Current methods often overlook brain activity heterogeneity across cognitive states.
• Characterizing brain-to-behavior relationships requires accounting for multi-state connectivity.

Purpose of the Study:

• To investigate the impact of multi-state functional connectivity on behavior.
• To develop a novel method for uncovering neurodevelopmental subtypes with distinct mechanisms.
• To analyze heterogeneous brain-to-behavior relationships in child development.

Main Methods:

• Proposed a nonparametric, Bayesian supervised heterogeneity analysis.
• Utilized stochastic block structures to identify network-based functional phenotypes.
• Developed a variational expectation-maximization algorithm for posterior computation.
• Integrated resting-state and task-related functional connectomes.

Main Results:

• Dissected heterogeneous effect mechanisms on children's fluid intelligence.
• Identified distinct functional network phenotypes (e.g., Fronto-parietal, Default Mode Networks).
• Demonstrated superior performance in uncovering brain-to-behavior relationships via simulations.

Conclusions:

• The proposed Bayesian method effectively uncovers neurodevelopmental subtypes and their distinct effect mechanisms.
• Accounting for multi-state functional connectivity is crucial for understanding brain-to-behavior relationships.
• This approach advances the analysis of brain development and child health.