Using phenomenological models for forecasting the 2015 Ebola challenge.

Area of Science:

• Epidemiology
• Mathematical Biology
• Infectious Disease Modeling

Background:

• Novel pathogens necessitate advanced mathematical modeling for epidemic forecasting.
• Accurate prediction of epidemic trajectory and size is crucial for public health response.

Purpose of the Study:

• To compare the performance of the logistic growth model and the generalized Richards model (GRM) for epidemic forecasting.
• To assess model accuracy in estimating the reproduction number, forecasting epidemic trajectory, and predicting final epidemic size.
• To evaluate models using synthetic data from the 2015 Ebola challenge.

Main Methods:

• Real-time forecasting using the logistic equation during the 2015 Ebola challenge.
• Post-challenge comparison of the logistic growth model and the generalized Richards model (GRM).
• Systematic assessment of model performance in estimating reproduction number, short-term incidence, and final epidemic size.

Main Results:

• The logistic equation consistently underestimated final epidemic size, peak timing, and peak case numbers.
• The GRM demonstrated superior performance in ascertaining final epidemic size with increasing data availability.
• GRM provided more accurate incidence forecasts across all scenarios, with lower root mean square error (RMSE) compared to the logistic model.
• Both models reasonably predicted the effective reproduction number, with GRM showing a slight advantage.

Conclusions:

• Models incorporating flexible early epidemic growth profiles are valuable for forecasting.
• The GRM's adaptability makes it particularly useful for evaluating infectious disease outbreaks using early-phase incidence data.
• These findings support the integration of advanced models into epidemic forecasting toolkits for improved accuracy and responsiveness.