Networks and the Epidemiology of Infectious Disease

The science of networks has revolutionised research into the dynamics of interacting elements. It could be argued that epidemiology in particular has embraced the potential of network theory more than any other discipline. Here we review the growing body of research concerning the spread of infectious diseases on networks, focusing on the interplay between network theory and epidemiology. The review is split into four main sections, which examine: the types of network relevant to epidemiology; the multitude of ways these networks can be characterised; the statistical methods that can be applied to infer the epidemiological parameters on a realised network; and finally simulation and analytical methods to determine epidemic dynamics on a given network. Given the breadth of areas covered and the ever-expanding number of publications, a comprehensive review of all work is impossible. Instead, we provide a personalised overview into the areas of network epidemiology that have seen the greatest progress in recent years or have the greatest potential to provide novel insights. As such, considerable importance is placed on analytical approaches and statistical methods which are both rapidly expanding fields. Throughout this review we restrict our attention to epidemiological issues.

💡 Research Summary

The paper provides a comprehensive review of how network science has been integrated into infectious‑disease epidemiology. It is organized into four main sections, each addressing a distinct aspect of the interdisciplinary field.

The first section classifies the types of networks that are relevant for modelling disease spread. Static graphs, which assume a fixed set of contacts, are useful for analytical tractability but fail to capture the temporal variability of human interactions. Time‑varying or temporal networks record when contacts occur and how long they last, allowing a more realistic representation of transmission pathways. Multilayer or multiplex networks incorporate several relational dimensions—such as household, workplace, and community contacts—within separate layers, thereby reflecting the complex, overlapping structures that characterize modern societies. Spatially embedded networks add geographic distance and mobility patterns, bridging local clustering with long‑range jumps. The authors argue that the choice of network representation should be driven by the research question, data availability, and the scale at which interventions are planned.

The second section surveys the quantitative descriptors used to characterise these networks. Degree distribution, especially heavy‑tailed (scale‑free) forms, highlights the presence of superspreaders and lowers epidemic thresholds. Clustering coefficients and triangle counts capture local cohesion, which can accelerate early spread but also raise the critical transmissibility needed for a global outbreak. Path‑length metrics (average shortest path, efficiency) relate directly to the speed of propagation. Community detection and modularity quantify the extent to which the network is compartmentalised, informing targeted vaccination or quarantine strategies. Centrality measures—betweenness, closeness, eigenvector, PageRank—identify nodes that are most influential for transmission, while core‑percolation analyses reveal the dense subgraph that often drives epidemic growth. The review emphasizes that no single metric suffices; a combination of structural indicators is required to predict epidemic outcomes accurately.

The third section focuses on statistical inference methods that estimate epidemiological parameters (e.g., transmission rate β, recovery rate γ) and, when necessary, the underlying network itself from partially observed data. Classical maximum‑likelihood estimation works well when the full contact network and infection times are known, but real‑world datasets are usually incomplete or noisy. Bayesian approaches—Markov chain Monte Carlo, variational inference, and Approximate Bayesian Computation (ABC)—allow incorporation of prior knowledge and quantification of uncertainty. Network‑specific models such as Exponential Random Graph Models (ERGM) and Stochastic Block Models (SBM) are described as tools for reconstructing missing edges or inferring latent community structure. Recent advances include graph‑neural‑network‑based inference, which can jointly learn transmission dynamics and network topology from time‑series case counts. The authors also discuss bias‑correction techniques for sampled networks and data‑augmentation strategies that mitigate the impact of unobserved contacts.

The fourth and final section compares analytical and simulation‑based techniques for predicting epidemic dynamics on a given network. Agent‑based simulations provide the most detailed representation, tracking each individual’s state and each edge’s activation, but they are computationally intensive for large populations. Percolation theory offers a powerful framework for deriving epidemic thresholds by treating disease spread as a bond‑percolation process on the graph. Mean‑field approximations, which assume homogenous mixing within degree classes, yield simple differential equations but ignore clustering and degree correlations. Pair‑approximation and higher‑order closures incorporate edge‑level correlations, improving accuracy at modest computational cost. Edge‑based compartmental models, which follow the probability that a randomly chosen edge transmits infection, have emerged as a versatile middle ground, capturing both heterogeneity and temporal dynamics. The review evaluates each method’s strengths, limitations, and suitability for policy‑relevant scenarios such as targeted immunisation, social‑distancing interventions, and real‑time outbreak forecasting.

Overall, the paper argues that network epidemiology has progressed from purely theoretical explorations to a mature toolbox that integrates high‑resolution contact data, sophisticated statistical inference, and scalable analytical models. It highlights the importance of matching the complexity of the network representation to the quality of available data and the specific public‑health question at hand. Future directions identified include real‑time inference on streaming mobility data, the fusion of multilayer network models with machine‑learning‑driven parameter estimation, and the development of adaptive control strategies that can be updated as network structures evolve during an outbreak.