Comparison of different models to analyze the number of patients in waiting-list

In this paper we have studied the behavior of the hospital activity analyzing the number of patients in waiting-list. This work considers different methods to determine the best structure of the model. The parametric models that have been analyzed are ARX, ARMAX and OE. The prediction error method has been used to estimate the model parameters. The models have been compared according with different error criteria: V, FPE and MAE. The autocorrelation and cross-correlation tests have been applied to validate the models. The best compromise between low MAE and low sensitivity of FPE has been obtained with ARMAX models. Simulations of three different surgeries with a mean absolute error less than 9 patients have been obtained using ARMAX models with a reduced number of parameters.