Modeling and forecasting hospital patient movements: Univariate and multiple time series approaches

The purpose of this paper is two-fold: it recognizes the importance of modeling and forecasting as a future-oriented decision-making process in the field of health care by proposing to model and forecast monthly patient movements in individual hospitals by means of the Box-Jenkins univariate and Tiao-Box multiple time series approaches; and it attempts to determine the choice between the univariate method and the multivariate approach in the case of hospital patient data. The forecasting performance of the resulting models is evaluated against the Holt-Winters exponential smoothing model using the mean squared error, mean absolute deviation, and mean absolute percentage error, for five non-overlapping time periods. The test results suggest that the proper choice of time series techniques to be applied to hospital admissions and discharges has an important and direct bearing on the reliability of forecasting results. Therefore, the practical implications and usefulness of time series techniques and models are discussed.