Network controllability measures of subnetworks: implications for neurosciences

Area of Science:

• Network Science
• Control Theory
• Neuroscience

Background:

• Network control theory applies control theory to understand complex systems.
• Controllability metrics quantify the functional role of network components.
• Neuroscience often lacks complete knowledge of all system variables.

Purpose of the Study:

• To investigate deviations in controllability metrics when applied to subnetworks versus full networks.
• To assess the translational value of controllability metrics in neuroscientific contexts with incomplete data.

Main Methods:

• Simulations using synthetic and structural MRI data.
• Systematic variation of network type, size, and edge density.
• Estimation of average and modal controllability on subnetworks and full networks.

Main Results:

• Controllability metrics (average and modal) are systematically over or underestimated in subnetworks.
• The deviation depends on the number of nodes in sub/full networks and edge density.
• These findings hold across various network types.

Conclusions:

• A systematic bias exists when applying controllability metrics to incomplete network data.
• Theoretical proof confirms generalization across network types.
• Understanding and mitigating this bias is crucial for accurate network analysis in neuroscience.