Incorporating Exogenous Factors in Adaptive Forecasting of Hospital Census

In this paper, we study the use of a recursive discounted least squares model for forecasting daily hospital census. We show, by means of a case study, how one can incorporate, with a minimum amount of cost and effort, institutional and exogenous factors explicitly into the model so as to enhance its forecast accuracy. These factors include holidays, capacity changes, and other events known to affect occupancies. The forecast modification comprises direct adjustments of model parameters, and judicious control of parameter updatings.