A Predictive Model for Hospital Patient Bed Forecasting

Session
Operational Performance Improvement

Author
Michael Meitzner
Principal Management Engineer, Henry Ford Health System

Description
Henry Ford Health System (HFHS) worked with Central Michigan University Research Corporation (CMURC) to investigate new approaches to forecasting patient-bed requirements by physician specialty. The association-sequence analysis methodology developed for retail marketing and store layout design provided insight into forecasting future hospital admission distribution.

Abstract
Henry Ford Health System (HFHS) worked with Central Michigan University Research Corporation (CMURC) to investigate new approaches to forecasting patient-bed requirements by physician specialty. The forecasts will be used to better plan short term operations and to help evaluate various capital investment plans for Henry Ford Hospital (HFH). The association-sequence analysis methodology developed for retail marketing and store layout design provided insight into the future admission distribution, given the recent outpatient and emergency room (O/E) visits within the HFHS network of clinics. The association-sequence analysis provides ?rules? to show the historical relationships between O/E visits by doctor specialty at a site and hospital admissions under the care of a doctor specialty; each rule is expressed in terms of frequency of occurrence, conditional probabilities and lag-time between the events. Analysis of total admissions by day provides holiday, seasonal and day-of-week factors to adjust the forecasts according to the specifics of each day in the forecast period. From the forecast of admissions by doctor specialty by day, historical length-of-stay and assignment to nurse station can be applied to obtain a bed-requirements profile.