Network inference with infection frequency matrix by improved Bayesian method.

Area of Science:

• Epidemiology
• Network Science
• Computational Biology

Background:

• Network inference often uses binary-state time series from epidemic models.
• Binary data lacks infection history, leading to inefficient and inaccurate network reconstruction.

Purpose of the Study:

• To develop a more efficient and accurate Bayesian network inference method.
• To improve the reconstruction of contact networks from epidemic data.

Main Methods:

• Proposed a Bayesian network inference approach.
• Converted binary-state time series into an infection frequency matrix.
• Jointly inferred contact network and transmission parameters.

Main Results:

• The infection frequency matrix representation reduced computational complexity.
• Achieved improved inference accuracy and fast, high-fidelity network reconstruction.
• Demonstrated a principled basis for optimizing intervention strategies.

Conclusions:

• The novel Bayesian method offers a significant improvement over traditional approaches.
• This approach enables efficient and accurate contact network reconstruction.
• Provides a foundation for optimizing interventions in various systems.