A new study in Applied Network Science examines the susceptible–infected–susceptible (SIS) epidemic process on a temporal multiplex network in which a persistent static layer is overlaid with an activity-driven layer of one-step contacts. The work, published on 22 August 2026 by Matin Marjani, Shirshendu Chatterjee, Sharmodeep Bhattacharyya and Caterina Scoglio, derives a compact analytical threshold for when an infection dies out or persists, and tests that formula against large-scale simulations on three standard random network families.
The model combines two sources of contact. The static layer is a fixed adjacency matrix whose spectral radius, denoted λ1, captures the connectivity of the backbone. The temporal layer follows an activity-driven construction: each node activates with probability a per time step, and when it activates it creates m one-step contacts. Because these contacts change from step to step, the authors replace the stochastic temporal contacts by their mean effect when analyzing the dynamics near the disease-free equilibrium.
Linearizing the SIS dynamics around the disease-free state under this mean-field approximation yields a threshold τc = 1/(λ1 + 2am). Below this value of the effective transmission parameter, the infection is predicted to go extinct; above it, the infection can persist. The formula makes explicit how the static backbone and the temporal activity layer jointly lower the epidemic threshold: increasing the number of contacts per activation or the activation probability shifts the extinction–persistence boundary in the same direction as strengthening the static network.
To evaluate the prediction, the team simulated the SIS process on Erdős–Rényi, Watts–Strogatz and Barabási–Albert networks, each with N = 10,000 nodes. The simulations displayed a clear extinction–persistence transition, and the location of that transition shifted systematically as the number of temporal contacts per activation increased. For the Erdős–Rényi and Watts–Strogatz backbones, the mean-field formula accurately tracked the empirical phase boundary observed in the simulations.
The scale-free case behaved differently. On the Barabási–Albert network, the spectral formula underestimated the empirical threshold by approximately 42% at N = 10,000. The authors attribute this discrepancy to degree heterogeneity and finite-size effects on scale-free networks, where a small number of highly connected nodes dominate the dynamics in ways that the leading eigenvalue alone does not fully capture at finite system size.
To close this gap, the researchers introduced an empirically calibrated correction factor and provided a corrected threshold formula that substantially improves agreement with simulations on the Barabási–Albert backbone. The correction is calibrated rather than derived from first principles, so its accuracy is tied to the network family and parameter regime in which it was fitted, and the authors present it as an improvement over the uncorrected spectral prediction rather than a universal replacement.
Beyond the threshold itself, the study reports three complementary diagnostics for locating the phase boundary: surviving-runs prevalence estimates, extinction probabilities, and early-time growth rates. These quantities probe the transition from different angles—how many infected remain among simulation runs that survive, how likely extinction is at a given parameter point, and how fast infections grow at early times—giving a more complete empirical picture of where the boundary lies than any single measure alone.
The practical significance of the result lies in its compactness. A single closed-form expression, requiring only the spectral radius of the static layer and the two activity-driven parameters a and m, predicts the epidemic threshold without requiring the full time-varying contact sequence. For networks whose degree distribution is relatively homogeneous, the formula appears reliable at the simulated scale; for heterogeneous scale-free backbones, the finite-N correction is needed for quantitative agreement.
The article is published open access, and the version shared on 22 August 2026 is an accepted-manuscript release that is citable and carries a permanent DOI, subject to further edits before the final Version of Record appears. The work sits within a broader literature on activity-driven network modeling of epidemic spread, including recent studies of concurrent epidemic strains on such networks, and contributes an analytically explicit threshold criterion for the combined static-plus-temporal setting that can be checked directly against simulation.
Subject of Research: Mean-field epidemic thresholds for SIS dynamics on temporal multiplex networks with activity-driven layers
Article Title: Epidemic mean-field thresholds of SIS dynamics on temporal multiplex networks with activity-driven layers
Article References: Marjani, M., Chatterjee, S., Bhattacharyya, S., & Scoglio, C. (2026). Epidemic mean-field thresholds of SIS dynamics on temporal multiplex networks with activity-driven layers. Applied Network Science. https://doi.org/10.1007/s41109-026-00824-0
Image Credits: AI Generated
DOI: 10.1007/s41109-026-00824-0
Keywords: SIS epidemic model, temporal multiplex networks, activity-driven networks, epidemic threshold, mean-field approximation, spectral radius, scale-free networks, Barabási–Albert network, Erdős–Rényi network, Watts–Strogatz network, finite-size effects, extinction–persistence transition
News Source: Kristina Jarvis. (November 7, 2026). Mean-field threshold formula tracks SIS epidemic extinction-persistence transition on temporal multiplex networks. Scienmag.



