Network inference from population-level observation of epidemics.

Area of Science:

• Epidemiology
• Network Science
• Computational Biology

Background:

• Inferring network structure from epidemic data is challenging, especially with limited population-level information.
• Real-world epidemic data often consists of aggregate counts of infected individuals at discrete time points.

Purpose of the Study:

• To develop a method for identifying the class of underlying networks (e.g., regular, Erdős-Rényi, Barabási-Albert) using only population-level epidemic data.
• To approximate continuous-time susceptible-infected-susceptible (SIS) models on networks using a Birth-and-Death process.

Main Methods:

• Utilized a Birth-and-Death process to model epidemic spread at the population level, with rates encoding network structure and disease dynamics.
• Generated network class-specific priors through extensive simulations on Regular, Erdős-Rényi, and Barabási-Albert networks.
• Employed Bayesian model selection to infer the most probable network class from a single epidemic realization.

Main Results:

• The proposed methodology successfully identified the underlying network class using limited, population-level epidemic data.
• Demonstrated effectiveness on both synthetic and real-world network data.
• The Birth-and-Death process effectively captured essential information about network structure from aggregate epidemic data.

Conclusions:

• It is feasible to infer network class from population-level epidemic data.
• The Birth-and-Death approximation provides a robust framework for analyzing epidemics on networks when detailed individual-level data is unavailable.
• This approach has implications for understanding disease spread and network properties in real-world scenarios.