Speaker
Description
Background and objectives
The transition to sentinel surveillance for SARS-CoV-2 has made it difficult to accurately determine the true scale of infections, underscoring the need for complementary indicators such as wastewater-based epidemiology (WBE). This study aims to estimate unobserved prevalence by developing a predictive framework that infers actual community-level infection scales directly from WBE data.
Methodology and results
A Survival Dynamical System (SDS) model was applied to national weekly SARS-CoV-2 wastewater concentrations and sentinel data in South Korea. The SDS integrates probabilistic distributions of infection and recovery times, enabling inference of continuous epidemic trajectories and estimation of true prevalence even when effective population size is unknown. Modelled prevalence was regressed against wastewater measurements, revealing a robust positive association. A Generalized Additive Model (GAM) with an Autoregressive (AR(1)) error term further captured nonlinear dynamics and time-series autocorrelation, yielding high predictive performance.
Implications
These findings demonstrate that wastewater viral concentrations serve as a reliable indicator of true infection prevalence. Integrating WBE with SDS modelling mitigates clinical underreporting, enabling timely estimation of epidemic intensity and transmission hotspots, and providing a robust complement to traditional public health surveillance.
