Urban-rural inequality in soil heavy metal health risks: Insights from Baoding, China

Soil heavy metal contamination poses serious health risks, but few studies have quantitatively assessed disparities in these risks between urban and rural populations. To address this gap, we introduce a novel framework integrating machine learning and spatially explicit risk models to assess individual- and population-level health risks from soil heavy metals in Baoding City, China. We used random forest models to predict high-resolution soil metal concentration maps, Positive Matrix Factorization for source apportionment, and spatial exposure models to estimate human health risks under multiple exposure pathways. This is the first study to combine high-resolution machine learning mapping, source apportionment, and multi-scale risk assessment in an urban-rural context. Key findings reveal risk contrasts: urban soils exhibited a 51 % higher ecological risk index than rural soils, reflecting concentrated pollution hotspots. However, individual-level risk assessments indicate that rural residents face 3 higher health hazards than urban residents. By contrast, aggregated non-carcinogenic and carcinogenic population risks were 1.8 and 1.7 times higher in urban areas. These contrasting results reveal an overlooked rural vulnerability at the individual scale versus greater aggregate risk in urban populations. Combining machine learning with spatially explicit risk modeling, our study quantifies previously undetected urban-rural health risk inequalities from soil contamination. This integrated approach advances scientific understanding of how urbanization shapes spatial health risk patterns and provides actionable insights for targeted environmental management to protect vulnerable communities, inform mitigation strategies, and identify priority intervention areas.