Analysis of a mathematical model for malaria using data-driven approach

Area of Science:

• Epidemiology
• Mathematical Modeling
• Environmental Science

Background:

• Malaria is a major global health burden causing millions of cases and deaths annually.
• Understanding disease transmission dynamics is crucial for effective public health interventions.

Purpose of the Study:

• To develop a novel framework for modeling malaria transmission dynamics.
• To integrate environmental factors (temperature, altitude) into a compartmental SIR-SI model.
• To enhance the realism and predictive accuracy of malaria spread models.

Main Methods:

• Developed a new transmission function incorporating temperature and altitude dependencies.
• Performed steady-state analysis to determine stability criteria for disease equilibria.
• Utilized a comparative learning framework with Artificial Neural Networks (ANNs), Recurrent Neural Networks (RNNs), and Physics-Informed Neural Networks (PINNs) for parameter estimation.
• Implemented Dynamic Mode Decomposition (DMD) to create a data-driven transmission risk index.

Main Results:

• Established stability criteria for disease-free and endemic equilibria.
• Physics-Informed Neural Networks (PINNs) demonstrated superior predictive performance by embedding epidemiological dynamics.
• A novel, interpretable transmission risk index was derived using DMD from infection data.

Conclusions:

• The novel framework enhances the realism of malaria transmission modeling by integrating environmental factors.
• Physics-constrained parameter inference using PINNs significantly improves predictive accuracy.
• The data-driven transmission risk index offers a valuable tool for real-time malaria risk assessment.