Abstract:
Methods aiming detection and inference of spatial clusters are of great relevance. Both for its applicability to public health problems, as well as for the effective scientific interest in the development of these methods. The main techniques are based on spatial scan statistics and many applications link this statistic in efficient optimization methods. Recently, regularity functions have been proposed to control the excessive freedom of the shape of spatial clusters whenever the spatial scan statistic is used. We present a novel scan statistic algorithm employing a function based on the geographical dispersion to penalize the presence of a huge gap between the areas in candidate clusters, the dispersion function. The proposed multi-objective scan maximize the spatial scan and minimize the dispersion function, using particle swarm optimization. We show that, compared to the non-connectivity regularity function, the multi-objective scan with dispersion function is faster and suited for the detection of moderately irregularly shaped clusters. An application is presented using comprehensive state-wide data for Chagas' disease in puerperal women in Minas Gerais state, Brazil. We show that, the multi-objective approach with the dispersion function is a promising algorithm for the detection of irregularly shaped clusters.
