The abstract is read at the publisher; the summary is JClub's.
Likelihood-based selection for the Weighted Stochastic Block Model identifies K=11 in adult human connectome networks and supports larger K around 15 in infants and toddlers.
- Why it matters: Community detection depends strongly on choosing the community count K, yet there is no standardized, principled guideline for resting-state functional connectivity networks.
- What they did: They benchmarked Weighted Stochastic Block Model, Spectral Clustering, and K-means on synthetic networks with ground truth and on adult plus infant neuroimaging cohorts, comparing multiple K-selection indices including a WSBM likelihood criterion with bootstrap confidence intervals.
- The result: On synthetic data, WSBM and Spectral Clustering recovered the true community number, while K-means did not; in adults, most indices lacked a unique optimum, but the likelihood criterion selected K=11, and in infants/toddlers it supported K around 15 with developmentally distinct mesoscale subdivisions.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.