Impact of behavioral heterogeneity on epidemic outcome and its mapping into effective network topologies

Human behavior plays a critical role in shaping epidemic trajectories. During health crises, people respond in diverse ways in terms of self-protection and adherence to recommended measures, largely reflecting differences in how individuals assess risk. This behavioral variability induces effective heterogeneity into key epidemic parameters, such as infectivity and susceptibility. We introduce a minimal extension of the susceptible-infected-removed~(SIR) model, denoted HeSIR, that captures these effects through a simple bimodal scheme, where individuals may have higher or lower transmission–related traits. We derive a closed-form expression for the epidemic threshold in terms of the model parameters, and the network’s degree distribution and homophily, defined as the tendency of like–risk individuals to preferentially interact. We identify a resurgence regime just beyond the classical threshold, where the number of infected individuals may initially decline before surging into large-scale transmission. Through simulations on homogeneous and heterogeneous network topologies we corroborate the analytical results and highlight how variations in susceptibility and infectivity influence the epidemic dynamics. We further show that, under suitable assumptions, the HeSIR model maps onto a standard SIR process on an appropriately modified contact network, providing a unified interpretation in terms of structural connectivity. Our findings quantify the effect of heterogeneous behavioral responses, especially in the presence of homophily, and caution against underestimating epidemic potential in fragmented populations, which may undermine timely containment efforts. The results also extend to heterogeneity arising from biological or other non-behavioral sources.

💡 Research Summary

The paper introduces a minimal extension of the classic susceptible‑infected‑removed (SIR) model, called HeSIR, to capture the impact of heterogeneous human behavior on epidemic dynamics. Individuals are divided into two behavioral classes: low‑risk (L) and high‑risk (H). High‑risk agents have increased susceptibility (α_S > 1) and increased infectivity (α_I > 1) relative to low‑risk agents. The model equations keep track of the fractions of susceptible, infected, and recovered individuals in each class, and define an “effective” infected fraction I_eff = I_L + α_I I_H that reflects the higher transmission potential of H‑infected persons.

To embed the model in a contact network, the authors use a degree‑corrected stochastic block model (DC‑SBM) with two communities, parameterized by the proportion p of high‑risk nodes and a homophily parameter h (0 ≤ h ≤ 1) that interpolates between random mixing (h = 0) and full assortative mixing (h = 1). The matrix of conditional connection probabilities π_ij incorporates both p and h, allowing the study of how preferential contacts among similar‑risk individuals shape spread.

The epidemic threshold is derived via heterogeneous percolation theory. Transmission probabilities across edges are ϕ_LL = 1 − e^{−β/γ}, ϕ_LH = 1 − e^{−α_I β/γ}, ϕ_HL = 1 − e^{−α_S β/γ}, and ϕ_HH = 1 − e^{−α_S α_I β/γ}. Fixed‑point equations for the probability that a randomly followed edge does not lead to the giant component are written in terms of the excess‑degree generating function g₁(x). Linearizing for small β/γ yields a closed‑form approximation:

 β/γ_c = ⟨k⟩ / (⟨k²⟩ − ⟨k⟩) · θ(p, h, α),

where α = α_S α_I and θ(p, h, α) = π_LL + α π_HH = (1 − h)(1 − p + α p) + h(α + 1). When α_S = α_I = 1, θ = 1 and the expression reduces to the classic threshold for homogeneous SIR on a configuration model. The factor θ captures how the joint presence of high‑risk individuals and their tendency to cluster (high h) amplifies the epidemic potential.

Early‑time analysis distinguishes two thresholds: one based on the growth of the total infected fraction I and another on the growth of the effective infected fraction I_eff. When both α_S and α_I exceed unity, the two thresholds differ, creating a “resurgence zone” where I may initially decline while I_eff continues to rise, eventually triggering a large outbreak. This phenomenon is analogous to a hidden reservoir of transmission that becomes dominant once enough high‑risk contacts have been activated. The authors note that a similar resurgence can arise purely from structural heterogeneity (e.g., heavy‑tailed degree distributions) even without behavioral heterogeneity, suggesting a unifying mechanism.

Extensive stochastic simulations validate the theory. Networks of N = 5 × 10⁴ nodes with average degree ⟨k⟩ = 20 are generated under various degree distributions (regular, Pareto with exponent a = 2 or 2.5) and homophily levels. Parameters α_S = 3, α_I = 2, and p = 0.4 are typical in the reported experiments. Simulations use an optimized Gillespie algorithm and seed 10⁻³ of the population uniformly at random. Results show:

1. Epidemic curves retain a qualitatively similar shape across all settings, but degree heterogeneity accelerates the peak due to superspreaders.
2. Increasing homophily concentrates high‑risk nodes, leading to earlier onset and higher peak prevalence.
3. The analytically predicted threshold (Eq. 4) aligns closely with the numerically observed transition points, identified both by the sharp rise in final epidemic size R_∞ and by the maximum of the variability measure VAR = ⟨R_∞²⟩ − ⟨R_∞⟩² / ⟨R_∞⟩. Small deviations in highly heterogeneous networks are attributed to finite‑size effects.

Finally, the authors demonstrate a mapping of HeSIR onto a standard SIR process on an “effective” directed SBM. In this mapping, the behavioral modifiers α_S and α_I are reinterpreted as adjustments to the out‑degree distribution (i.e., the number of contacts) rather than to per‑contact transmission probability. Consequently, the same epidemic threshold emerges, but the heterogeneity is now encoded in network topology. This duality provides a conceptual bridge: behavioral heterogeneity can be equivalently viewed as structural heterogeneity, facilitating the design of interventions that target either contact patterns or behavioral compliance.

Overall, the study offers a unified analytical framework that integrates behavioral, biological, and structural sources of heterogeneity. It quantifies how risk‑related traits, their clustering, and network degree variability jointly shape epidemic thresholds, early dynamics, and the possibility of hidden resurgence. The findings have practical implications for risk assessment, the timing of non‑pharmaceutical interventions, and the design of policies that consider both individual behavior and the underlying contact network.