Supplementary Datasets for: Predictive Modeling of Adult Arsenic Health Risks near an Abandoned Gold Mine in Loei, Thailand

This dataset contains the supplementary materials for the manuscript, "Predictive Modeling of the Spatial Gradient in Adult Arsenic Health Risks: An Integrated Environmental and Survey Data Approach Near an Abandoned Gold Mine, Loei Province, Thailand." Abstract Chronic arsenic exposure from legacy mining presents a severe, long-term public health threat. This study developed an integrated predictive model to identify the key environmental, dietary, sociodemographic, and behavioral determinants of arsenic health risks for adults residing in a spatial gradient near a former gold mine in Loei Province, Thailand. A cross-sectional study of 371 adults was conducted, integrating a quantitative Human Health Risk Assessment (HHRA) with primary survey data. Multivariate statistical modeling, including logistic regression and the machine learning algorithm XGBoost, was used to analyze the non-cancer Hazard Index (HI), Incremental Lifetime Cancer Risk (LCR), and predictors for diagnosed arsenicosis. The HHRA revealed unacceptable health risks. Non-cancer HI values reached 7.0, and LCRs ranged from 2.0×10−3 to 4.4×10−3, exceeding the U.S. EPA's upper-limit threshold by 20 to 44 times. Residential proximity to the mine was the dominant predictor of overall risk. The model identified two critical exposure pathways: consumption of locally grown rice, which significantly predicted higher exposure scores (p<0.001), and the source of water for all daily household activities—not only drinking—which was the strongest predictor of diagnosed arsenicosis (OR for interaction with distance = 1.94, p<0.001). While socioeconomic status was not a direct predictor, factors such as household structure and alcohol consumption significantly modulated risk. Paradoxically, higher risk awareness trended with increased odds of diagnosed arsenicosis (p=0.077). These findings demonstrate an urgent public health crisis driven by a complex interplay of where people live, non-drinking water exposures, and food chain contamination. These findings mandate that interventions move beyond conventional safe drinking water programs to address the entire domestic water system and the local food supply, underpinned by robust environmental remediation and targeted public health strategies