Predicting Short-term Hospital Inpatient Bed Needs for Up to Four Days in Advance

The Scarborough Hospital embarked a corporate initiative called Bed Mapping, which maps a better path for inpatient care and has patients in the right bed. A tool is needed to perform short-term predictions of inpatient bed needs. This thesis provides an inpatient demand predictive tool that uses historical, scheduled surgery, and current occupancy data to predict inpatient bed needs for up to four days in advance. It uses staff generated Estimated Discharge Date (EDD), Length of Stay (LOS) distributions, and a Monte Carlo simulation model by Liu [36] to make predictions. First, the tool was tested using retrospective CIHI expected patient LOS data to calculate EDD, and showed that the tool’s error is within 12 beds (5.52%) on day-1 and increases as we go further from day-1. The tool was further tested and showed that if EDD’s were accurate then the tool’s error is within two beds (0.88%) on day-1.M.A.S