A sticky weighted regression model for time-varying resting-state brain connectivity estimation.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Network Science

Background:

• Current brain connectivity models often assume static networks, which contradicts the brain's inherent nonstationarity.
• Understanding temporal dynamics is crucial for accurately modeling brain function.

Purpose of the Study:

• To develop and validate a time-varying model for brain connectivity networks.
• To investigate temporal and spatial differences in brain connectivity associated with Parkinson's disease (PD).

Main Methods:

• A novel time-varying model incorporating fused least absolute shrinkage and selection operator (LASSO) for abrupt changes and weighted regression for smooth changes.
• Simulations to compare the proposed method against static and weighted time-varying regression models.
• Application to resting-state functional magnetic resonance imaging (fMRI) data from PD and control subjects.

Main Results:

• The proposed method demonstrated improved accuracy in estimating time-dependent connectivity patterns.
• Significantly different temporal and spatial connectivity patterns were identified in PD subjects compared to controls.
• PD subjects exhibited reduced network variability over time.

Conclusions:

• The developed time-varying model accurately captures dynamic brain connectivity.
• Reduced network variability in PD may reflect impaired cognitive flexibility.
• Temporal dynamic properties of brain connectivity could serve as potential biomarkers for PD.