牛海沣 麻豆a片 讲师

摘要:

Urban health disparities impact over half the world's population and pose a critical barrier to achieving ‘Good Health and Well-Being’, a key Sustainable Development Goal. Despite extensive evidence on individual health determinants, existing urban health studies often rely on linear modelling frameworks and relatively coarse spatial units, limiting their ability to capture non-linear effects, threshold behaviour and spatial heterogeneity in how social and environmental factors jointly shape local physical health and mental well-being outcomes. This study addresses this gap by applying a spatially explicit, explainable machine-learning framework that combines XGBoost with SHAP analysis to examine health determinants at a fine spatial grid scale. The results reveal socio-economic factors—such as unemployment, housing barriers and low income—as the most significant drivers, par-ticularly influencing mental well-being. Gender composition and elderly population ratios are moderate predictors, while built environment factors, such as population density and points of interest, show weaker but distinct associations. Threshold effects, including the negative impacts of extreme heat and housing barriers, further underscore the complexity of urban health determi-nants. The findings highlight the need for systems-based, context-sensitive interventions to address these challenges and provide actionable insights to inform equitable urban health policies, advancing global well-being.

关键词:

explainable machine learning | geospatial analysis | health determinants | non-linear effects | urban health | XGBoost

点击此处查看原文