Analyzing the Application of Spatial Autocorrelation Measures in Geographic and Epidemiological Statistics

Posted: Feb 01, 2008

• Luna Nguyen
• Maria Flores
• Michael Wright

Abstract

Spatial autocorrelation represents a fundamental concept in geographical analysis, describing the degree to which similar values cluster together in space. The application of spatial autocorrelation measures in epidemiological statistics has traditionally followed established methodologies developed for geographic data analysis. However, the unique characteristics of disease transmission patterns, including temporal dynamics, population mobility, and environmental factors, necessitate specialized approaches that conventional spatial statistics may not adequately address. This research addresses critical gaps in current methodologies by developing an integrated framework that accounts for the complex interplay between spatial structure and disease dynamics. The conventional application of spatial autocorrelation in epidemiology has largely focused on global measures such as Moran's I and local indicators like LISA (Local Indicators of Spatial Association). While these methods provide valuable insights, they often assume stationarity in spatial processes and rely on predetermined neighborhood structures that may not reflect the actual mechanisms of disease transmission. Our research challenges these assumptions by introducing adaptive spatial weighting functions that incorporate disease-specific transmission parameters, population density gradients, and mobility patterns. This study is motivated by three primary research questions that have received limited attention in the existing literature. First, how can spatial autocorrelation measures be adapted to account for the dynamic nature of disease transmission across different spatial scales? Second, what methodological innovations can improve the detection of non-stationary spatial patterns in epidemiological data? Third, how can hybrid approaches combining traditional spatial statistics with machine learning techniques enhance our understanding of disease clustering mechanisms?