Multitask deep learning for the emulation and calibration of an agent-based malaria transmission model

Area of Science:

• Computational epidemiology
• Machine learning applications in public health
• Disease modeling

Background:

• Agent-based models (ABMs) are crucial for malaria transmission research but are computationally intensive to calibrate.
• Accurate calibration is essential for reliable disease dynamics understanding and intervention planning.

Purpose of the Study:

• To develop and validate a multitask deep learning approach for emulating and calibrating a complex agent-based model of malaria transmission.
• To demonstrate the potential of machine learning-guided emulator design for complex scientific processes.

Main Methods:

• Trained a neural network emulator on extensive simulations from the EMOD malaria model.
• Captured relationships between immunological parameters and epidemiological outcomes (incidence, prevalence).
• Utilized the emulator with parameter estimation techniques to calibrate the model to reference data.

Main Results:

• The neural network emulator effectively captured complex relationships within the malaria transmission model.
• The approach successfully calibrated the agent-based model to reference data from eight sub-Saharan African sites.
• Demonstrated the feasibility of using machine learning for efficient model calibration.

Conclusions:

• Machine learning-guided emulator design offers a powerful approach for accelerating the calibration of complex scientific models.
• This methodology can significantly improve the efficiency of malaria transmission modeling and intervention planning.
• The study highlights the potential of integrating AI with epidemiological modeling for public health advancements.