Spatial Analysis of Schizophrenia Prevalence Using a Multiobjective Evolutionary Algorithm

A group of geographically close spatial units that shows a similar prevalence pattern – significantly highis called a hotspot and, when it is low or drain-like –outlet effect, this area is called a coldspot. Both areas can be identified through the evaluation of the degree of agreement between Local Indicators of Spatial Aggregation (LISA) and Bayesian Conditional Autorregresive (CAR) scores. A Multi Objective Evolutionary Algorithm (MOEA) has been designed and tested to evaluate this degree of agreement and QQ-Plots are used to identify hot and cold-spots. The MOEA uses four different strategies to evaluate the fitness function including a fuzzy approach and the standard SPEA2. Using this methodology the spatial distribution throughout the time span (years 2004, 2006, 2007 and 2008) of schizophrenia prevalence has been analysed in 770 municipalities in Andalusia (southern region in Spain). Our procedure is a robust method to identify both hot and cold-spots in the space and can be useful to organize the spatial distribution of health-care services.