Short-term Covid-19 forecast for latecomers.

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Despite vaccination, several countries face significant COVID-19 waves.
• Accurate short-term forecasting of cases and fatalities is essential for healthcare management.

Purpose of the Study:

• To develop a reliable statistical method for real-time, short-term forecasting of COVID-19 cases and deaths.
• To specifically address the needs of 'latecomer' countries in their pandemic response.

Main Methods:

• Utilized a penalized LASSO regression model incorporating an error correction mechanism.
• Modeled latecomer countries based on countries at similar pandemic stages previously.
• Employed an adaptive rolling-window scheme for forecasting.

Main Results:

• The proposed methodology demonstrated strong performance across 45 countries.
• Detailed analysis for Brazil, Chile, Mexico, and Portugal confirmed its effectiveness.
• The model significantly outperformed alternative forecasting methods.

Conclusions:

• The penalized LASSO regression with error correction provides accurate short-term forecasts for COVID-19.
• This approach aids in better short-run healthcare system management.
• The framework is adaptable for regional forecasting and other infectious diseases.