Forecasting local hospital bed demand for COVID-19 using on-request simulations.

Area of Science:

• Epidemiology
• Healthcare Management
• Public Health

Background:

• Infectious disease epidemics necessitate accurate hospital bed demand forecasting to prevent healthcare system overload.
• Existing forecasting methods may lack the granularity required for effective individual hospital planning.

Purpose of the Study:

• To develop an intuitive online tool for individual hospitals to forecast COVID-19 bed demand.
• To assess the impact of geographical catchment area selection on forecast accuracy and precision.

Main Methods:

• Development of an online forecasting tool utilizing local incidence, vaccination, and bed occupancy data.
• Customizable geographical resolutions for hospital catchment areas.
• Performance evaluation through ICU bed occupancy forecasting for German university hospitals and regions.

Main Results:

• Optimal forecasting accuracy achieved when the catchment area aligns with the hospital's local catchment.
• Expanding catchment area improved precision at the cost of accuracy.
• Forecasting performance decreased during epidemic turning points.
• Incorporating variants of concern had minimal impact on overall bed occupancy forecasts.

Conclusions:

• Local data utilization is significant for accurate epidemic forecasts.
• Hospital-specific catchment area forecasts offer a superior balance of accuracy and precision compared to national or state-level data.
• These forecasts provide actionable insights for hospital planning, including adjustments to elective surgeries for capacity management.