Data-Guided Sectorisation by Spatio-Temporal Autocorrelation Approach for Epidemiological Surveillance

This paper presents a methodology for the automatic spatial division of a region using spatial and spatio-temporal autocorrelation indices for epidemiological surveillance. Traditional methods often fail to reveal spatial and temporal relationships between disease outbreaks. Our approach utilizes global (Moran and Geary indices) and local (Local Gi* from Getis-Ord) autocorrelation measures to identify clusters, trends, and anomalies in disease data. The framework was applied to tuberculosis surveillance in Oran, Algeria, using geographical and temporal data. The results show significant clusters of high tuberculosis incidence, with the Local Gi* revealing critical hotspots. Spatio-temporal analysis with the LISA index captured evolving disease patterns, providing insights for targeted public health interventions. This approach enhances the understanding of disease dynamics, enabling public health officials to focus resources on high-risk areas and improve disease control efforts. Future work includes refining temporal analysis and applying the methodology to other diseases