Feed

Compartmental models are essential for studying host-pathogen dynamics, evaluating intervention effectiveness, and predicting infection trends. However, the utility of these models for testing competing hypotheses is often overlooked. To address this, we propose a new model-based hypothesis testing (MBHT) approach, which uses compartmental models to evaluate the hypotheses in epidemiology. In our case, using the COVID-19 pandemic as a case study, we formulate hypotheses of SARS-CoV-2 mutation and construct a transmission model to test them. In addition to analyzing steady-state stability, deriving the basic reproduction number, and identifying a backward bifurcation, the model is fitted to seven peaks of U.S. COVID-19 data, each corresponding to periods of viral mutation and morbidity peaks. The estimated posterior probabilities reveal that Short-term within host selection primarily shaped mutations during the early pandemic stages, followed by immune selection driven by natural and vaccine-induced immunity. In later stages, mutations aligned with vaccination-induced virulence and transmission-virulence correlation, while the declining virulence and immune selection partially explained the final stages of SARS-CoV-2 mutation. In conclusion, model-based hypothesis testing offers a powerful yet underutilized approach to uncovering drivers of viral mutation and gaining deeper insights into pathogen evolution during outbreaks.