Advancing hybrid spatial dynamic modeling for urban sustainable transitions

Guiding urban sustainable transitions requires advanced tools to anticipate and shape future spatial change. Spatial Dynamic Modeling (SDM) is pivotal, and its hybridization with Machine Learning (ML) has become a dominant trend for capturing complex, non-linear urban dynamics. This study aims to comprehensively examine the state of the art in hybrid SDM oriented toward urban sustainable transitions, with implications for planning support systems (PSS). Accordingly, we review and synthesize the methodological landscape and thematic applications of hybrid SDM models from the early 2010s to the present. Our analysis reveals that the field is at a critical juncture, shifting from a tool for pattern prediction toward a platform for transition support, yet constrained by several fundamental deficits. First, methodologically, while ML has substantially improved the estimation of transition probabilities in raster-based frameworks, the field still lags in adopting vector-based architectures necessary for simulating fine-grained decision-making mechanisms. Second, regarding evaluation, despite reported performance gains, the absence of standardized benchmarks and validation protocols renders cross-study comparison difficult. Third, thematically, a misalignment exists in research prioritizing physical morphological assessment over human-centric dimensions such as spatial equity and system-level network resilience. To operationalize the transition support capacity, we propose a future research agenda: hybrid SDM should pivot from algorithmic novelty to framework innovation, establish rigorous open-science benchmarking standards, and evolve from structural assessment to system performance simulation. By addressing these gaps, hybrid SDM can transform into a robust and collaborative platform for navigating equitable and resilient urban futures.