Epidemiological patterns and trends of respiratory syncytial virus in Guangdong Province, China, 2019-2023.

Background: Respiratory syncytial virus (RSV) is a major cause of acute respiratory tract infections worldwide. The COVID-19 pandemic and associated non-pharmaceutical interventions (NPIs) substantially altered the transmission dynamics of respiratory viruses. However, post-pandemic changes in RSV epidemiology and their associations with meteorological factors remain insufficiently characterized in subtropical regions of China. Method: Based on 38,541 RSV test records collected in Guangdong Province from July 1, 2019 to July 2, 2023, the epidemic characteristics were analyzed across the pre-pandemic, pandemic, and post-pandemic phases. The shortest main epidemic period in each season was identified using a sliding window approach. Monthly percentage change was used to quantify temporal variations across epidemic seasons. Meteorological data were integrated to assess temporal trends and correlations. A Distributed Lag Non-linear Model (DLNM) was applied to quantify non-linear exposure-response relationships and lag effects of meteorological factors. An XGBoost model combined with SHAP was used for feature attribution and performance evaluation. Results: RSV positivity rates exhibited pronounced seasonal and interannual variability throughout the study period. Epidemic periods ranged from 7 to 10 months and frequently extended beyond the traditional winter-spring season. Overall positivity rates increased from 8.19% in the pre-pandemic phase to 13.84% during the pandemic phase and 19.33% in the post-pandemic phase. Significant differences were primarily observed among children younger than 6 years. Multiple meteorological variables were associated with RSV activity. DLNM analyses identified non-linear threshold effects and short-term lag effects (0-3 weeks), with elevated RSV risk associated with specific ranges of temperature, dew point temperature, and precipitation. XGBoost models captured major temporal patterns in RSV activity, with lagged RSV indicators contributing most strongly to predictive performance, whereas meteorological variables provided modest incremental predictive value. Conclusion: RSV activity in Guangdong during 2019-2023 was characterized by prolonged epidemic periods and altered seasonal patterns following the COVID-19 pandemic. Meteorological factors demonstrated non-linear and lagged associations with RSV activity, highlighting their potential role in shaping transmission dynamics. The combined DLNM and XGBoost-SHAP framework provided complementary insights into climate-related risk patterns and temporal forecasting of RSV activity. Continued surveillance and external validation are needed to determine whether the observed seasonal changes represent a stable long-term pattern and to improve the generalizability of predictive models.