A Bayesian Framework for the Network Analysis of Transmission Dynamics in Infectious Disease.

Area of Science:

• Epidemiology
• Computational Biology
• Network Science

Background:

• Understanding infectious disease transmission is vital for public health.
• Traditional models often oversimplify complex human contact patterns.
• Accurate transmission network reconstruction is challenging.

Purpose of the Study:

• To develop an extended Bayesian framework integrating genomic, temporal, and network data for precise transmission network reconstruction.
• To account for social and spatial proximity in transmission probability calculations.
• To enhance inference sensitivity using a hypothesis testing procedure.

Main Methods:

• Developed an extended Bayesian framework incorporating network structure as a prior.
• Integrated genomic, temporal, and social network data.
• Employed a hypothesis testing procedure optimized via constrained likelihood estimation.
• Applied Exponential Random Graph Models (ERGM) for network analysis.

Main Results:

• Network-informed models significantly outperform non-network-informed models, especially with limited genetic data.
• The model effectively resolved transmission ambiguities in a tuberculosis (TB) dataset from Kampala, Uganda.
• Transmission was found to be more probable via weak social ties than within dense clusters.

Conclusions:

• The integrative Bayesian framework provides a robust tool for epidemiological analysis.
• Network-informed models offer superior accuracy in reconstructing transmission dynamics.
• Findings support targeted public health interventions and decision-making.