Multimodal Brain Connectivity, Baseline Characteristics, and Executive Function in Preterm Infants

Overview<br>This dataset comprises multiple components that capture non-negative matrix factorization (NMF) outputs, graph-theoretical metrics, clinical confounders, and executive function (EF) scores in a cohort of very preterm infants (≤32 weeks’ gestation). The data are organized into several files, each reflecting a different level of analysis or data type.1. All SC FeaturesContent: Comprehensive structural connectivity (SC) metrics (e.g., average path length (ASP), betweenness centrality (BC), clustering coefficient (CC), cost (CO), local efficiency (LE), and regional degree (RD)) derived from the diffusion MRI (dMRI) data.Purpose: Enables replication or secondary analysis of the graph-theoretical metrics before feature selection and robust regression modeling.2. All FC FeaturesContent: Comprehensive functional connectivity (FC) metrics for the same graph-theoretical measures (ASP, BC, CC, CO, LE, RD) derived from resting-state fMRI data.Purpose: Similarly provides the complete set of functional connectivity metrics for analyses parallel to the SC data.3. Baseline CharacteristicsContent: Clinical and demographic data for each infant, including key confounders such as gestational age (GA), antenatal corticosteroids, retinopathy of prematurity, bronchopulmonary dysplasia, maternal age, etc.Purpose: Allows researchers to control for or stratify by these confounders when examining the relationship between brain connectivity and developmental outcomes.4. EF ScoreContent: The Minnesota Executive Function Scale (MEFS) standard scores measured at three years corrected age (CA) for each participant.Purpose: Serves as the dependent variable in analyses linking SC/FC metrics to EF outcomes in preterm infants.5. FC Top Vars &amp; SC Top VarsContent: The subsets of functional (FC) and structural (SC) variables that were selected via LASSO for robust regression modeling (e.g., top variables contributing significantly to EF prediction).Purpose: Facilitates focus on the key connectivity metrics (from NMF-derived modes of inter-subject covariation) that emerged as predictive of EF, rather than the full set of metrics.6. Top Variable Scores for Both FC and SCContent: Individual-level data reflecting the highest-contributing FC and SC graph metrics across the main categories (ASP, BC, CC, CO, LE, RD).Purpose: Allows a direct comparison of the crucial graph-theoretical measures— both structural and functional—that appear most relevant to EF outcomes.Usage NotesNMF Outputs: These are “modes of inter-subject covariation,” capturing overlapping patterns of connectivity across participants, not canonical resting-state networks.Confounder Control: Data on baseline characteristics enable robust multivariable modeling.EF Score: The age-corrected MEFS standard scores are suitable for early childhood EF assessment, offering a unidimensional measure of executive function.Reproducibility: Researchers can replicate or extend the original analyses by merging the “All SC Features” / “All FC Features” with the “Baseline Characteristics” and “EF Score,” then focusing on “Top Vars” if they aim to confirm or refine the robust regression results.