Phase transitions and emergent behavioral coordination in coupled awareness–epidemic dynamics on complex networks: An agent-based approach with autonomous decision-making

During an epidemic, individual decisions to mask, distance, or vaccinate continuously reshape transmission, yet how this collective behavioral response interacts with contact-network structure, which determines whether voluntary behavior alone can contain an outbreak, remains poorly understood. Coupled awareness–epidemic models on multiplex networks have shown that information and disease co-evolve and can shift the epidemic threshold, but they almost universally treat individuals as passive nodes governed by fixed Markov rules, leaving autonomous, multi-dimensional decision-making outside their scope and rarely linking behavior to an analytical threshold or to network topology. To close this gap, a hybrid agent-based and graph-theoretic framework is proposed in which each node is an autonomous agent that decides on masking, distancing, and vaccination through sigmoid functions driven by compliance, dynamic awareness, and local epidemic state, with awareness co-evolving with both the epidemic and the contact topology. A heterogeneous mean-field analysis yields the effective reproduction number in closed form as the product of transmissibility, the network heterogeneity ratio, and a behavioral reduction factor, defining a topology-dependent threshold under which scale-free networks are far more vulnerable than homogeneous ones at equal average degree. Simulations across six scenarios and four topologies, validated on three empirical SocioPatterns networks and extended to ten thousand nodes, show that the advantage of autonomous behavior over exogenous intervention is conditional: pronounced on homogeneous structures but diminishing on scale-free networks and against aggressive structural targeting. The results identify contact-network heterogeneity as the decisive factor setting the limits of behavior-based epidemic control.