A comprehensive modelling framework to forecast the demand for all hospital services

Area of Science:

• Healthcare Management
• Operations Research
• Health Informatics

Background:

• Hospitals face intense demand pressure, leading to resource shortages.
• Accurate patient demand forecasting is essential for effective hospital management.
• This need applies across all hospital services, including inpatient, outpatient, and A&E.

Purpose of the Study:

• Develop a comprehensive forecasting model for all acute hospital services.
• Support hospital management in better demand management and future planning.
• Improve preparedness for fluctuations in patient service utilization.

Main Methods:

• Utilized national episodes statistics dataset for demand capture.
• Applied four forecasting methodologies: ARIMA, exponential smoothing, multiple linear regression, and STLF.
• STLF was applied for the first time in healthcare demand forecasting.

Main Results:

• Selected 64 optimal forecasting models from 760 developed models.
• Forecasted demand for 38 outpatient, 25 inpatient specialties, and A&E services.
• Identified optimal forecasting periods (daily, weekly, monthly) for different specialties.

Conclusions:

• No single forecasting method or period suits all hospital services; 'one size does not fit all'.
• Seasonal and Trend decomposition using Loess (STLF) outperformed traditional methods for several specialties.
• Accurate demand forecasting enables effective planning, resource allocation, budget management, and risk reduction.