"Back to the future" projections for COVID-19 surges

Area of Science:

• Epidemiology
• Mathematical Modeling
• Infectious Disease Dynamics

Background:

• COVID-19 pandemic necessitates accurate forecasting of infection surges.
• Traditional epidemiological models often struggle with early prediction of inflection points.
• International data sharing can provide valuable insights for localized outbreak management.

Purpose of the Study:

• To introduce and validate a novel forecasting method called back-to-the-future (BTF) projections.
• To assess the accuracy of BTF projections in predicting COVID-19 surges across diverse global settings.
• To compare the predictive performance of BTF projections against traditional modeling approaches.

Main Methods:

• Development of a matching scheme for asynchronous time series data.
• Integration of time series matching with a response coaching SIR (Susceptible-Infectious-Recovered) model.
• Application of the BTF projection method to COVID-19 data from 12 countries across populated continents.

Main Results:

• BTF projections accurately predicted future COVID-19 surges prior to the daily infection curve's inflection point.
• The method demonstrated superior performance compared to traditional approaches, which often predicted no future surges.
• BTF projections were unable to predict surges driven by the emergence of new variants, as anticipated.

Conclusions:

• Back-to-the-future projections offer a powerful tool for anticipating COVID-19 surges using international data.
• This approach enhances epidemic preparedness by providing earlier warnings than conventional models.
• The method's limitations highlight the ongoing challenge of predicting outbreaks influenced by viral evolution.