Novel Eigen space method for multiple Spatiotemporal rare diseases clusters detection: a case study of waterborne disease.

The development of robust and efficient analytical tools for informed decision making, mainly in epidemiological contexts, remains a persistent challenge. This study presents an enhanced algorithm designed to accurately detect vulnerable spatiotemporal hotspots associated with unexpected disease outbreaks. We introduce an improved novel Multi-EigenSpot algorithm by systematically integrating the functionalities of both EigenSpot and its Multi-HotSpot extension. The EigenSpot algorithm effectively identifies single spatiotemporal clusters, it is unable to detect multiple hotspots. The Multi-EigenSpot algorithm overcomes this limitation through an iterative process of cluster detection and removal. However, challenges persist regarding computational efficiency and sensitivity in identifying rare clusters. To address these limitations, we propose an efficient Novel Multi-EigenSpot algorithm. This method is designed to detect multiple irregularly shaped, rare spatiotemporal clusters with significantly improved computational performance. Furthermore, the proposed algorithm integrates heatmap visualizations to enhance the interpretability of detected clusters. We evaluated our method using monthly waterborne disease surveillance data from Khyber Pakhtunkhwa, Pakistan (January - December 2024), comparing its performance against both the original EigenSpot and Multi-EigenSpot algorithms. Empirical results demonstrate the proposed algorithm's superior performance in accurately identifying multiple spatiotemporal clusters. Beyond public health surveillance, this algorithm is readily adaptable to diverse domains, including crime analysis, environmental hazard detection, and other applications requiring spatiotemporal clustering.