A0202
Title: A functional model-adjusted spatial scan statistic
Authors: Michael Genin - University of Lille (France) [presenting]
Mohamed Salem Ahmed - University of Lille (France)
Abstract: A functional model-adjusted spatial scan statistic is introduced to adjust the detection of clusters on longitudinal confounding factors indexed in space. An approach based on generalized linear functional models is used to construct this spatial scan statistic, with longitudinal confounding factors being considered as functional covariates. A general framework is proposed for various probability models and application to the Poisson model shows that this method is equivalent to a classical statistical spatial scan statistic considering an underlying population adjusted for covariates. Through a simulation study, we show that this method has a better quality of adjustment than other methods based on univariate and multivariate models. The proposed method is illustrated using premature mortality data in France during the period from 1998 to 2013, considering the quarterly unemployment rate as a longitudinal confounding factor.
