Residential Mobility: Implications for the Accuracy of Disease Cluster Detection

University Of North Carolina At Charlotte, Charlotte NC

Investigators

Abstract

Public health organizations are increasingly relying on space-time clustering techniques to target place-based health initiatives such as prevention. However, the methodological techniques benefit from refinement. In this proposal, the researchers examine the impact of commonly cited concerns about the validity of space-time cluster detection while developing methods and practices to resolve or reduce their effects on cluster analytics. Specifically, the researchers first study the impact of residential mobility on the validity of space-time clustering. Second, the researchers estimate the level of granularity at which residential histories should be studied before clusters become unnoticed. A third goal of the project is an evaluation of how space-time relationships constructed from patient mobility can affect the relative risk of existing space-time clusters. The interdisciplinary project contributes to research in health geography and space-time analytics. The study integrates new sources of individual movement data while creating methodological tools to incorporate those data into existing workflows. These methods and tools contribute to public health efforts to respond to disease outbreaks in the future. Key findings and recommendations are shared with local, regional, and federal stakeholders. Multiple students are involved in the project, contributing to the education and training of early-career social scientists. The complexity of human mobility, disease dynamics, and privacy concerns are persistent challenges to the validity of cluster detection methods. The opportunity to test the consequences of these assumptions is now possible through improved access to electronic health records, individual residential histories, accurate linkage algorithms, and advanced geospatial technologies. In this project, the researchers examine how heterogeneous spatial and temporal resolution affects space-time cluster detection accuracy. The project also examines how the integration of residential mobility may mitigate space-time uncertainty, with the aim of discerning the level of detail needed to maximize the detection of spatiotemporal clustering in the data. Methodologically, the project uses simulated space-time trajectories to test clustering algorithms, simultaneously evaluating how scaling issues and other perturbations affect the validity of clustering tests. The generalizability of the approach is examined via an analysis of two large empirical datasets that include detailed detail on the mobility of patients after diagnosis. The project informs research design considerations by highlighting the extent to which residential mobility should be monitored for accurate inferences. In addition to applications to disease detection, the methods extend to other areas of research that use clustering methods, such as spatial ecology and animal movement, transportation and urban dynamics, and criminology. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.