Normative Modeling of Static and Dynamic Functional Connectivity.

Area of Science:

• Neuroscience
• Statistical Modeling
• Neuroimaging Analysis

Background:

• Methodological heterogeneity in functional neuroimaging confounds subject-specific effects with study variance.
• Normative modeling offers a statistical framework to address and parameterize variability in datasets.
• Harmonizing fragmented legacy datasets is crucial for robust individual-level brain modeling.

Purpose of the Study:

• To evaluate the efficacy of normative modeling in harmonizing heterogeneous functional MRI datasets from multiple cohorts.
• To construct a lifespan normative chart for brain functional connectivity without intensive data reprocessing.
• To investigate age-related trajectories of static and dynamic functional connectivity.

Main Methods:

• Utilized a multi-level statistical model (generalized additive model) incorporating study-level random effects.
• Harmonized open-access functional MRI data from seven large cohorts (N = 4705) with heterogeneous processing pipelines.
• Drove normative trajectories of static and dynamic functional connectivity from a well-calibrated lifespan model.

Main Results:

• Successfully harmonized fragmented legacy neuroimaging data, creating a lifespan normative chart.
• Identified distinct age-related trajectories for static and dynamic functional connectivity.
• Static connectivity showed a monotonic age-related decline, while dynamic connectivity exhibited complex age-dependent changes, including mid-adulthood metastability.

Conclusions:

• Normative modeling provides a scalable statistical framework for analyzing functional brain organization across diverse datasets.
• This approach circumvents the need for massive data homogenization while preserving sensitivity to individual variability and temporal dynamics.
• The findings highlight a key decoupling between static and dynamic connectivity, offering new insights into brain aging and development.