Temporal exponential random graph models of longitudinal brain networks after stroke.

Area of Science:

• Neuroscience
• Network Science
• Computational Biology

Background:

• Stroke-induced brain plasticity involves widespread functional reorganization beyond the immediate injury site.
• The precise local mechanisms driving these large-scale network changes over time remain poorly understood.
• Traditional analyses often treat time as static observations rather than a dynamic variable.

Purpose of the Study:

• To investigate the role of temporal connection motifs in explaining brain network reorganization after stroke.
• To develop and validate a statistical framework for analyzing time-varying brain networks.
• To link dynamic network changes to long-term patient behavioral outcomes.

Main Methods:

• Utilized temporal exponential random graph models (tERGMs) to analyze dynamic functional brain networks.
• Implemented temporal motifs (T and E) as parameters within tERGMs.
• Validated the model on synthetic networks and applied it to human stroke data from 2 weeks to 1 year post-stroke.

Main Results:

• tERGMs successfully captured brain network changes occurring after stroke over a 1-year period.
• Identified specific temporal signatures: within-hemisphere segregation (T) and between-hemisphere integration (E).
• Interhemispheric temporal edges (E) significantly correlated with chronic language and visual outcomes.

Conclusions:

• Time-varying network properties are crucial for understanding complex dynamic systems like the brain after stroke.
• Temporal connection motifs provide novel insights into the mechanisms of brain reorganization and functional recovery.
• Dynamic network analysis offers predictive potential for patient outcomes following stroke.