Fig. 3: Generative network modeling schematic.
From: Premature birth changes wiring constraints in neonatal structural brain networks
a Diagram depicting how models form over time by adding connections in the network, based on a trade-off between parameterized cost (di,j) and value (ki,j), to mimic network growth. Cost is a static metric defined by the Euclidean distances between nodes. Value is a topological metric that changes every time a new connection is added. The model is generated by iteratively by calculating the wiring probability of all potential connections and selecting a single connection based on the probability score. b The process of creating simulations starts by performing a grid search to identify the parameter combinations for each run, for example, 10,000 runs = 10,000 parameter combinations. Each participant’s generative models were created with the same parameter combinations; however, the models grew to the size of that individual’s network based on the total connections present in the participant’s observed network. Then, for each participant, we compared their finished models to their observed network and identified the single best-fit model for that individual. Node plots in both panels are adapted from AAL90 atlas in Shi F, et al. (2011) Infant Brain Atlases from Neonates to 1- and 2-Year-Olds. PLoS ONE 6(4): e18746. doi:10.1371/ journal.pone.0018746.
