Synthetic method of analogues for emerging infectious disease forecasting

Area of Science:

• Epidemiology
• Computational Biology
• Public Health

Background:

• The Method of Analogues (MOA) is a popular non-parametric approach for infectious disease forecasting.
• MOA relies on matching current time series data to historical data, which is a limitation when historical data is sparse, as seen during the COVID-19 pandemic.

Purpose of the Study:

• To introduce the Synthetic Method of Analogues (sMOA), a novel forecasting method designed to overcome the limitations of MOA in data-scarce scenarios.
• To evaluate the performance of sMOA against state-of-the-art infectious disease forecasting models.
• To present a new uncertainty quantification methodology for emerging epidemics.

Main Methods:

• sMOA generates forecasts by matching ongoing time series data to a library of synthetic disease trend data, bypassing the need for extensive historical data.
• The study compared sMOA's performance against models in the COVID-19 Forecasting Hub using Mean Absolute Error and Weighted Interval Score.
• A novel uncertainty quantification method was developed and applied to emerging epidemic scenarios.

Main Results:

• sMOA demonstrated competitive performance compared to existing state-of-the-art infectious disease forecasting models.
• sMOA outperformed 78% of models in the COVID-19 Forecasting Hub based on averaged Mean Absolute Error.
• sMOA outperformed 76% of models in the COVID-19 Forecasting Hub based on averaged Weighted Interval Score.

Conclusions:

• sMOA offers a viable alternative for infectious disease forecasting, particularly in situations with limited historical data, such as the onset of new epidemics.
• The developed uncertainty quantification methodology is crucial for public health decision-making during novel pandemics.
• Versatile forecasting approaches that do not depend on historical data are essential for improving global health preparedness.