Academia Sinica Logo

COVID-19 reinfections continue due to antigenic drift, waning immunity, and behavioural changes, but how these processes interact with transmission dynamics is not fully understood. Our group develops transmission models and machine learning approaches that integrate diverse experimental data, including viral traits derived from sequences and serological responses from vaccinated individuals, to better understand variant fitness and epidemic dynamics. Our models revealed a rugged fitness landscape shaped by ACE2 binding and effective immunity, with discrete peaks for each variant of concern. Using data from multiple countries, we found that socioeconomic inequalities can exacerbate the impact of immune‑escape variants. In Hong Kong, increased vaccination reduced infection rates during co‑circulation, while successive immune‑escape variants spread more by out‑competing earlier strains, underscoring stronger selection pressure. Understanding the determinants of viral fitness can improve vaccine strain selection and vaccination strategy.