EpiRank: Modeling Bidirectional Disease Spread in Asymmetric Commuting Networks

Area of Science:

• Epidemiology
• Network Science
• Computational Biology

Background:

• Commuting networks exhibit asymmetrical flows, facilitating infectious disease transmission.
• Understanding spatial epidemic risk requires analyzing bidirectional commuting patterns.

Purpose of the Study:

• To introduce EpiRank, a novel algorithm for assessing infection risk within commuting networks.
• To evaluate EpiRank's performance against existing network analysis methods and real-world disease data.

Main Methods:

• Developed EpiRank using a Markov chain model and PageRank-like algorithm.
• Integrated forward and backward commuting movements using a daytime parameter.
• Utilized Taiwan's 2000 census data for network construction and risk mapping.

Main Results:

• The bidirectional movement model in EpiRank accurately captured spatial epidemic risk.
• EpiRank outperformed PageRank and HITS algorithms in predicting disease distribution.
• Backward movement was found to be more critical than forward movement for disease diffusion analysis.

Conclusions:

• EpiRank provides a superior method for analyzing disease diffusion in commuting networks.
• The bidirectional nature of commuting is crucial for accurate epidemic risk assessment.
• EpiRank offers a valuable alternative for public health surveillance and intervention planning.