Assessing the Spatiotemporal Characteristics, Factor Importance, and Health Impacts of Air Pollution in Seoul by Integrating Machine Learning into Land-Use Regression Modeling at High Spatiotemporal Resolutions

• DOI: 10.1021/acs.est.2c03027
• Publication Date: 2023-01-11T21:16:34Z
• Authors: Yue Li; Tageui Hong; Yefu Gu; Zhiyuan Li; Tao Huang; Harry Fung Lee; Yeonsook Heo; Steve H. L. Yim
• Keywords: Predictive modelling; Health effect

Previous studies have characterized spatial patterns of air pollution with land-use regression (LUR) models. However, the spatiotemporal characteristics pollution, contribution various factors to them, and resultant health impacts yet be evaluated comprehensively. This study integrates machine learning (random forest) into LUR modeling (LURF) intensive evaluations develop high resolution prediction models estimate daily diurnal PM2.5 NO2 in Seoul, South Korea, at 500 m for a year (2019) then evaluate driving quantify premature mortality. Our results show that incorporating random forest algorithm our model improves performance. Meteorological conditions great influence on models, while play important roles assessment using dynamic population data estimates when combined, causes total 11,183 (95% CI: 5837–16,354) mortalities Seoul 2019, which 64.9% are due PM2.5, remaining attributable NO2. The pollution-attributable largely caused by cardiovascular diseases including stroke. pinpoints significant variations impact providing essential epidemiological research quality management.