Using data-driven agent-based models for forecasting emerging infectious diseases.

Area of Science:

• Epidemiology
• Computational modeling
• Infectious disease dynamics

Background:

• Forecasting emerging infectious diseases presents significant challenges, including poor data quality and limited understanding of disease transmission.
• Policy decisions during epidemics require timely and reliable predictions, which are often hindered by complex, dynamic factors.

Purpose of the Study:

• To describe an agent-based model (ABM) framework developed for forecasting the 2014-2015 Ebola epidemic in Liberia.
• To evaluate the model's performance and utility in a real-world epidemic scenario and forecasting challenge.

Main Methods:

• Development of a detailed agent-based model integrating disease dynamics and social behavior.
• Calibration of the model using available data from the Ebola epidemic in Liberia.
• Summarization and analysis of forecast performance across various scenarios during the Ebola forecasting challenge.

Main Results:

• The agent-based model provided a framework for integrating diverse data sources to model epidemic spread.
• The model's performance was assessed against actual epidemic trajectories during the forecasting challenge.
• Key components and calibration strategies for the model were detailed.

Conclusions:

• Data-driven computational models, such as the described ABM, offer a valuable approach for understanding and forecasting epidemic dynamics.
• The framework can be refined and adapted for preparedness and response to future infectious disease outbreaks.
• Lessons learned from the Ebola epidemic forecasting challenge provide insights for improving epidemic modeling and response strategies.