Fig. 3: Development of the structural hubs and modules from birth to childhood.

A To delineate the maturation of brain hubs and modules, we used a sliding window approach to generate averaged brain networks and topological metrics at different age points for 100 subgroups. B Structural network hubs from birth to age 8 (left hemisphere). The distributions of regional efficiency (white to red) and brain hubs (the top 10% of nodes in terms of nodal efficiency, within black lines) at 12 representative age points ranging from 0.5 years to 7.5 years are shown. For each point, the regional efficiency was averaged across individuals. For detail views, see Fig. S2. C A k-means clustering analysis was performed on pairwise correlations (left panel) between nodal efficiency after de-meaning across age subgroups, the best fit was found with two clusters (Fig S3), with an age boundary before and after 2.5 years of age. The distributions of the hub regions in each cluster and the consistent/inconsistent regions between clusters are plotted (right panel; orange colour: hubs in cluster 1; red colour: hubs in cluster 2; green colour: common hubs in two clusters. D Age-related changes (uncorrected p values, two-tailed) in global modularity and the participant coefficient of the brain structural connectome. The scatter plot shows the fitted developmental curves with the shaded area indicating the 95% confidence interval around the fitted mean. E A multilayer module detection approach was implemented by an age-ordered conjunction of all averaged matrices in the age subgroups. F Cortical module affiliations at 12 representative age points (top view). For detail views, see Fig. S5. G High modular flexibility was identified in the frontal association cortex during development (blue to green represent values from low to high). Source data are provided as a Source Data file.