Vigilância espaço-temporal com superfícies acumuladas

Abstract

Mapping crime and disease events provides information about the spatial and the temporal pattern of these events. To guide actions based on such information, it is necessary to use astatistical method to identify when there is a change in the pattern of events. We developed aspace-time prospective surveillance method when the data are point events, monitoring if there is an emerging cluster. At each new event, a local Knox score is calculated and spatially spread toform a stochastic surface. The surfaces are accumulated sequentially until it overcomes a specified threshold, when an alarm goes o, identifying the region of the probable cluster. The advantages are to require little prior knowledge from the user and to provide a way to identify the locations of the possible clusters, through the visualization of the cumulative surface. A simulation study is presented for dierent scenarios and a dataset of meningitis cases in Belo Horizonte was monitored.