A Deep Dynamic Causal Learning Model to Study Changes in Dynamic Effective Connectivity During Brain Development.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Developmental Neuroscience

Background:

• Dynamic effective connectivity (dEC) offers insights into brain development but remains largely unexplored due to limitations in existing methods.
• Current approaches often fail to capture the time-varying nature of brain information transmission.

Purpose of the Study:

• To develop and validate a novel deep dynamic causal learning model for capturing brain dEC.
• To investigate age-related differences in dEC patterns using real-world neuroimaging data.

Main Methods:

• A deep dynamic causal learning model integrating a dynamic causal learner and discriminator was proposed.
• The model processes spatio-temporal fMRI data to identify and validate time-varying causal relationships.

Main Results:

• The model demonstrated superior accuracy in identifying dynamic causalities compared to existing methods on simulated data.
• Application to the Philadelphia Neurodevelopmental Cohort revealed distinct dEC network patterns across age groups.
• Brain dEC networks in young adults showed greater stability than in children, with significant differences in information transfer.

Conclusions:

• The study highlights the brain's developmental shift from undifferentiated to specialized networks with age.
• This transition correlates with enhanced cognitive and information processing capabilities.
• The proposed model accurately detects dEC and characterizes its age-related evolution.