Abstract
BACKGROUND: Respiratory syncytial virus (RSV) poses a significant disease burden among children <5 worldwide. Yet systematic analyses of how complex environmental factors are associated with RSV transmission are still lacking in many countries. METHODS: We introduced a novel 3-stage, data-driven framework to assess the impacts of environmental factors, including meteorological conditions, air pollutants, and extreme weather, on RSV infections from a spatiotemporal perspective. It includes (1) spatiotemporal patterns of RSV transmission; (2) a hierarchical model (HSDLNM) to examine associations between environmental factors and RSV transmission, estimating relative risks (RRs) and 95% confidence intervals; and (3) an interpretable machine learning model, Gaussian Process Boosting, to predict RSV infections using historical environmental data. We validated the applicability of the proposed framework in Japan. RESULTS: Weekly data on the number of newly lab-confirmed RSV-positive cases, meteorological factors, and air pollutants were collected from 47 Japanese prefectures (2013-2019). We identified the meteorological thresholds strongly linked to elevated RSV infections, particularly weekly average temperature <10°C (RR, 1.10) or >20°C (RR, 1.13) and weekly average relative humidity <60% (RR, 1.04) or >70% (RR, 1.06). Short-term exposure to particulate matter of 2.5 μm (PM2.5) is associated with elevated infection risk. Additionally, historical environmental data aid in forecasting RSV activities in Japan. CONCLUSIONS: This study presents the relationships between environmental factors and RSV infections in Japan. Our framework could be applied to areas with similar RSV seasonality to further understand environmental impacts regionally. This research helps inform policy decisions on RSV prophylaxis strategies, supporting cost-effective measures for controlling and preventing early transmission.