Estimating the two consecutive epidemic waves of SARS-CoV-2 Omicron in Shenzhen, China from November 2022 to July 2023: a modeling study based on multi-source surveillance and mobility data

Unlike many countries that have experienced multiple COVID-19 waves since January 2020, China’s stringent measures left most of the population without natural immunity to SARS-CoV-2. After lifting controls, China experienced two distinct Omicron waves from November 2022 to July 2023. However, no reliable study has yet elucidated the transmission dynamics of these two consecutive Omicron waves in China’s megacities, nor the phenomenon of reinfection due to immune escape. To address this gap, this study proposes a hybrid epidemic modeling framework based on multi-source surveillance and mobility data, including nucleic acid tests, wastewater surveillance, case reports from Notifiable Infectious Diseases Surveillance System of China (NIDSS), and intra-urban travel intensity data. In this hybrid modeling framework, a four-stage compartmental model stratified by age is developed, integrating human mobility and Omicron reinfection mechanisms. This model is further corrected by an agent-based model to address the overestimation of infections by the compartmental model, forming a comprehensive hybrid framework. Based on the simulation results, several new findings are drawn. The attack rate of the first wave in Shenzhen was 88.5% (95% confidence interval (CI): 72.1%-99.6%), lower than other models’ predictions. The peak of the second wave occurred on May 18, 2023, with a higher reinfection rate compared to those observed in other countries and regions. The effective reproduction number ( $R_{t}$ ) for the first wave peaked at 5.44 (95% CI: 5.26-5.48), while for the second wave, the initial $R_{t}$ was 1.28 (95% CI: 1.27-1.29). The first infections provide a 0.549 (95% CI: 0.544-0.554) protective effect against XBB reinfection within six months. In conlusion, this study presents an advanced modeling framework for accurately assessing epidemic spread in urban environments using multi-source surveillance data.