PROJECT SUMMARY The striking disparities in disease dynamics and health outcomes between various socio-demographic groups observed during the SARS-CoV-2 pandemic have highlighted the critical need for improved modeling approaches that help us understand and predict such disparities. Most modeling studies currently produce insights at spatial scales that are too coarse (e.g., counties, states) in scale, and are usually insensitive to local socio-demographic variations in disease dynamics; this limits their practical value for informing public health planning in local communities. What is needed is a set of standardized methods to efficiently create fine- grained models capable of exposing and capturing key spatial features and socio-demographic factors that impact transmission dynamics and disease outcomes, and that can reveal how different socio-demographic sub-groups may experience disparate disease outcomes. Here we propose a novel SSCIMA (Social-Spatial Clustering, Interconnection, and Movement Analysis) modeling approach to efficiently expose linkages between local mobility, socio-demographic composition, and evolving disease surveillance and to optimize the construction of meta-population disease models that can make more accurate disease forecasts at the scales of census blocks (i.e., local neighborhoods). We will use SARS-CoV-2 and simulated data sets to drive the design and rigorous testing of generalized methods and software that will be useful for future pandemic preparedness. In Aim 1, we develop statistical methods that ingest disease data, mobility patterns, and socio- demographic statistics at the scale of census blocks and use these data to determine the features that most strongly explain patterns of local and regional transmission dynamics as well as disease outcomes (e.g., hospitalization rates). In Aim 2, we develop efficient methods leveraging the linkages revealed under Aim 1 to fit meta-population models to sparse data at the scale of census blocks, integrating mobility data, and socio- demographic features to yield high-fidelity meta-population models structured directly based on evolving observed patterns of disease dynamics. Analyses driven by simulated and real data will reveal the potential for SSCIMA-driven configuration of meta-population models to improve local forecast accuracy; and we will also produce freely available software and a cloud-based modeling portal to allow exploration and testing of our method and tools. In Aim 3, we will focus on dissemination and education, developing a new educational module for deployment within SHERC’s existing outreach infrastructure, as well as a half-day training workshop for the modeling community to learn about, engage with, and provide feedback on our technique and tools. We expect the SCCIMA approach to enable more rapid, spatially refined, and equity-focused modeling efforts that will better equip us for future epidemic events.