The human brain is organized as a dynamic network system whose functional architecture changes across time, individuals, development, and disease. Although functional magnetic resonance imaging has enabled large-scale mapping of intrinsic brain networks, many existing approaches emphasize either temporal connectivity dynamics or static spatial organization, leaving important aspects of spatiotemporal network variability under-characterized. This dissertation develops and applies computational frameworks for modeling brain network dynamics across two complementary contexts: network dysfunction in schizophrenia and normative functional brain maturation during early infancy. First, this work examines dynamic default mode network chronnectomics in schizophrenia using an explainable feature-learning framework. Dynamic functional network connectivity states were identified across two independent schizophrenia cohorts, and state occupancy was used to characterize group differences and symptom associations. An iterative feature- importance procedure revealed reproducible default mode network edges contributing to dynamic state organization across datasets, with symptom-relevant effects concentrated in anterior–posterior default mode interactions. These findings support the view of schizophre- nia as a disorder of altered large-scale network dynamics and demonstrate the value of interpretable dynamic feature learning for linking network states to clinical phenotypes. Second, this dissertation characterizes early postnatal functional brain network develop- ment using longitudinal natural-sleep resting-state fMRI data acquired from neurotypical infants between birth and six months of age. Group ICA and individualized network esti- mation were used to identify reproducible infant functional networks, and multiple spatial metrics were developed to quantify network maturation. Initial analyses examined age- related changes in network-averaged spatial similarity, network engagement range, network strength, network size, and network center of mass, revealing network-specific patterns of spatial refinement, expansion, and reorganization. Subsequent nonlinear trajectory analyses showed that infant networks do not mature uniformly; instead, different systems exhibit dis- tinct developmental shapes, suggesting both coordinated maturation and network-specific differentiation during the first half-year of life. Finally, this dissertation introduces a multidimensional framework for infant brain net- work maturation by integrating five previously defined spatial metrics with two new mea- sures: fractal dimension, indexing network spatial complexity, and network center displace- ment, indexing deviation from group-level network location. Using age-specific normaliza- tion, generalized additive models, principal component analysis, inter-network coupling, and joint ICA, this framework maps both within-network maturation and system-level coordina- tion across 15 infant functional networks. Results show that infant functional brain devel- opment is characterized by multiple nonlinear modes of spatial maturation and coordinated coupling, rather than a single global developmental process. Together, these studies advance spatiotemporal modeling of functional brain networks by providing interpretable tools for quantifying dynamic dysfunction in schizophrenia and multidimensional maturation in infancy. This work contributes a unified perspective in which brain networks are understood not only by where they are located or how strongly they connect, but by how their spatial organization, temporal dynamics, and inter-network relationships evolve across development and disease.