Abstract:Background: Environmental exposure to toxic metals is an important risk factor to human health. Traditional methods have examined associations between a health endpoint and exposure to heavy metals by either univariate or multiple regression. In the setting of ubiquitous heterogeneous environmental exposures, statistical methods that incorporate mixed exposures are increasingly relevant and may provide new insight into the association between metal exposure and important cardiovascular, renal and respiratory outcomes. Objective: The objective of this study was to classify the population of National Health and Nutrition Examination Survey (NHANES) into different exposure subgroups using modern unsupervised clustering methods based on lead, cadmium, mercury, and arsenic measured in urine or whole blood, and to assess the association between the identified exposure groups and twelve important health endpoints. Methods: We analyzed a sub-cohort of 9662 subjects participating in the 6 cycles (2003-2004 to 2013-2014) of NHANES study. The urine levels of 3 heavy metals (total arsenic, lead, cadmium) and blood levels of 3 heavy metals (lead, cadmium and mercury) were analyzed using a two-step approach. In the first step, we stratified the population into subgroups using unsupervised clustering (k-medoids) based on levels of metals either in urine or in blood. Then, we examine the association between 12 health endpoints and identified exposure subgroups while controlling for age, sex, race/ethnicity, education, smoking status, BMI, and urinary creatinine. Results: The k-medoids algorithm clustered NHANES population into 2 groups based on either blood or urinary levels of heavy metals. The concentrations of all the three heavy metals were significantly different between the identified groups in blood (p < 2.2e-16) or in urine (p = 0). The group with higher concentrations was defined as the "high-exposure" group, while the group with lower concentrations was defined as "low-exposure" group. Association analysis with health outcomes suggested that the high-exposure group according to either blood or urinary metal levels had significantly higher total mortality (1.63-1.64 times higher, p < 0.0001), mortality caused by malignant neoplasms (2.05-2.62 times higher, p < 0.0002), Gamma-glutamyl transferase (GGT) (1.03-1.05 times higher, p < 0.0001). In addition, the high-exposure group based on blood levels was also significantly associated with SBP, death related to hypertension, heart disease and chronic lower respiratory disease, while the high-exposure group based on urinary concentrations had higher mortality related to nephritis. Conclusions: We proposed an unsupervised clustering method to stratify the population into high- and low-exposure groups based on the co-exposure of heavy metals. The high-exposure groups, characterized by higher metal concentrations, had significant higher GGT, SBP, DBP, and mortality rates suggesting the detrimental effects of exposure to these heavy metals. The stratification of the NHANES population based on exposure patterns provides an informative method to study the impact of metal exposures on health outcomes.

Pesticides, cancer, and oxidative stress: an application of machine learning to NHANES data

Application and progress of machine learning in pesticide hazard and risk assessment

Quantitative identification of the co-exposure effects of e-waste pollutants on human oxidative stress by explainable machine learning

Environmental chemical exposures and a machine learning-based model for predicting hypertension in NHANES 2003–2016

Using Four Machine Learning Methods to Analyze the Association Between Polycyclic Aromatic Hydrocarbons and Visual Impairment in American Adults: Evidence from NHANES

Building a predictive model for hypertension related to environmental chemicals using machine learning

Development of a Machine-Learning Model to Identify the Impacts of Pesticides Characteristics on Soil Microbial Communities from High-Throughput Sequencing Data.

Unveiling the development trends of environmental and human health concerns for pesticides: Integrating an intelligent approach and data mining across diverse databases

A prospective cohort study of exposure to household pesticide with cardiovascular diseases mortality in older adults

Identification for heavy metals exposure on osteoarthritis among aging people and Machine learning for prediction: A study based on NHANES 2011-2020

Development of Interpretable Machine Learning Models Associated with Environmental Chemicals to Predict All-Cause and Specific-Cause Mortality：a Longitudinal Study Based on NHANES

Exposure to phenols, chlorophenol pesticides, phthalate and PAHs and mortality risk: A prospective study based on 6 rounds of NHANES

Applying machine learning to construct an association model for lung cancer and environmental hormone high‐risk factors and nursing assessment reconstruction

Unraveling the pesticide-diabetes connection: A case-cohort study integrating Mendelian randomization analysis with a focus on physical activity's mitigating effect

Machine learning-driven identification of early-life air toxic combinations associated with childhood asthma outcomes

Advancing source apportionment of soil potentially toxic elements using a hybrid model: a case study in urban parks, Beijing, China

Linking industrial emissions and dietary exposure to human burdens of polychlorinated naphthalenes

Machine Learning-Based Toxicological Modeling for Screening Environmental Obesogens

Stratification of Population in NHANES 2009–2014 Based on Exposure Pattern of Lead, Cadmium, Mercury, and Arsenic and Their Association with Cardiovascular, Renal and Respiratory Outcomes

Evaluating county-level lung cancer incidence from environmental radiation exposure, PM2.5, and other exposures with regression and machine learning models

Potential lethality of organochlorine pesticides: Inducing fatality through inflammatory responses in the organism