Bed allocation techniques based on census data

This paper considers the problem of allocating a number of beds to the different medical and surgical specialties in a hospital. This task is complicated by the fact that patterns of patient arrivals differ among specialties, that the scheduling of medical procedures varies over a week, and that the demand for various medical services can show seasonality. We develop a time series model using hourly census data to make good decisions regarding the size of each unit while minimizing data collection and modeling efforts. This model is simple enough to be used widely and, we suggest, represents an improvement versus methods currently employed by hospitals. Results are presented from an application to data collected at Northside Hospital, Atlanta, GA.