Facing rising demand pressures and costs, hospitals dynamically make managerial decisions based on patient inventory (census) estimates. Specifically, hospital staffing, resource, and financial decisions are made according to forecasted inpatient demand. Hospitals make decisions on different timeframes including daily, weekly, and monthly wherein seasonal patterns occur. Previous models for inpatient demand prediction are subjective or require extensive inputs such as medical patient attributes. These models can be inaccurate and difficult to implement. This research presents a Discrete Time Markov Chain (DTMC) methodology, based solely on historical local patient census, which predicts both short and long term inpatient census. Our DTMC model is superior to past inpatient models since it tests for and incorporates census seasonality, requires only easily obtainable data to populate, can be implemented as a spreadsheet application, and can be smoothed by regression to predict unobserved census levels.
展开▼
