[The follow-up and monitoring of forecasts: the case of asthma epidemics in Barcelona].

In this work we discuss several procedures of evaluation of time series models and analyze practical aspects of the management of a data base relative to a health information system devoted to the epidemiological surveillance and the detection of unusual values that could correspond to epidemic events. The analyzed statistical information corresponds to the time series of the number of emergency room admissions for asthma in several hospitals of Barcelona city, between the first of January of 1985 and the seventh of september of 1987. Among the three analyzed methods, the one based on the correlations of the one-step-ahead residuals (the second one) is the most specific and, furthermore, it doesn't require the previous definition of an epidemic day. In addition, this second procedure has less analytical and computational difficulty.