| Abstract: |
| We develop a unified multiscale modeling framework that integrates within-host viral kinetics with population-level epidemic dynamics, combining analytical and agent-based approaches. Motivated by SARS-CoV-2 variant evolution, we first construct a mechanistic SEIIRS-renewal model in which infection-age-dependent infectiousness is derived from viral load trajectories through a nonlinear Hill-type mapping. This formulation enables rigorous analytical characterization of epidemic thresholds, including a kernel-based basic reproduction number, Euler--Lotka growth rates, and explicit conditions for backward bifurcation induced by waning immunity and reinfection.
To complement this theoretical framework, we develop a multi-scale agent-based model that embeds empirically inferred viral kinetics into network-based transmission dynamics. By mapping individual viral load trajectories to time-varying transmission probabilities, the model captures heterogeneous contact structures and stochastic transmission processes, allowing for realistic simulation of variant-specific epidemic patterns.
Our results demonstrate that differences in viral replication and clearance fundamentally reshape epidemic dynamics, even under identical reproduction numbers. Fast-replicating variants generate earlier and sharper epidemic peaks, whereas slower viral dynamics lead to delayed but more prolonged outbreaks. The integration of within-host kinetics further produces heterogeneous transmission patterns and heavy-ailed secondary infection distributions without imposing ad hoc assumptions.
Together, the proposed framework establishes a coherent bridge from biological mechanisms to population-level outcomes, providing both analytical insight and computational tools for understanding variant-specific transmission dynamics. This unified multiscale approach offers a flexible foundation for epidemic forecasting, risk assessment, and timing-sensitive intervention design in emerging infectious diseases. |
|