A Multilevel Machine Learning Framework for Mapping and Predicting Diffuse and Point-Source Heavy Metal Contamination in Surface Soils

Soil heavy metal pollution threatens agricultural safety and human health, with Cd exceeding standards being the most common problem in contaminated farmland. The development of hyperspectral remote sensing technology has provided a novel methodology of quickly and non-destructively monitoring heavy metal contamination in soil. This study aims to explore the potential of an interpretable Stacking ensemble learning model for the estimation of soil Cd contamination in farmland hyperspectral data. We assume that this method can improve the modeling accuracy. We chose Zhangjiagang City, Jiangsu Province, China, as the study area. We gathered soil samples from wheat fields and analyzed soil spectral data and Cd level in the lab. First, we pre-processed the spectra utilizing fractional-order derivative (FOD) and standard normal variate (SNV) transforms to highlight the spectral features. Second, we applied the competitive adaptive reweighted sampling (CARS) feature selection algorithm to identify the significant wavelengths correlated with soil Cd content. Then, we constructed and compared the estimation accuracy of multiple machine learning models and a Stacking ensemble learning method and utilized the Optuna method for hyperparameter optimization. Ultimately, the SHAP method was used to shed light on the model’s decision-making process. The results show that (1) FOD can further highlight the spectral features, thereby strengthening the correlation between soil Cd content and wavelength; (2) the CARS algorithm extracted 3.4–6.8% of the feature wavelengths from the full spectrum, and most of them were the wavelengths with high correlation with soil Cd; (3) the optimal estimation precision was achieved using the FOD1.5-SNV spectral pre-processing combined with the Stacking model (R2 = 0.77, RMSE = 0.05 mg/kg, RPD = 2.07), and the model effectively quantitatively estimated soil Cd contamination; and (4) SHAP further revealed the contribution of each base model and characteristic wavelengths in the Stacking modeling process. This research confirms the advantages of the interpretable Stacking model in hyperspectral estimation of Cd contamination in farmland wheat soil. Furthermore, it offers a foundational reference for the future implementation of quantitative and non-destructive regional monitoring of heavy metal contamination in farmland soil.

Source apportionment and risk assessment of heavy metal contamination in soils surrounding the tailings based on copper isotopes.

Monitoring heavy metal(loid) concentrations in soils of industrially contaminated sites using machine learning models.

Estimation of heavy metals using deep neural network with visible and infrared spectroscopy of soil

Anthropogenic activities such as mining, improper waste disposal, and agricultural practices have contributed significantly to heavy metal contamination in soils, posing substantial ecological and public health risks. This study assessed the levels of heavy metal contamination in soils across four locations: Lambangudu, Maijankasa, Gidan Kara Bodel, and Quentin Gate in Gashaka Local Government Area (LGA), Taraba State, Nigeria. Using the Geo-accumulation Index (Igeo), soil samples were analysed for nine heavy metals: cadmium (Cd), lead (Pb), cobalt (Co), copper (Cu), chromium (Cr), zinc (Zn), nickel (Ni), manganese (Mn), and iron (Fe). The results revealed that Mn, Pb, and Fe exhibited extreme contamination (Igeo > 6) across all sites, indicating strong anthropogenic influence, likely from artisanal mining and unregulated waste disposal. Ni and Co also recorded high contamination levels, falling within the heavily to extremely contaminated class (Igeo Class 5), while Zn showed moderate contamination. Cd, Cu, and Cr remained within uncontaminated levels (Igeo Class 0), suggesting limited anthropogenic inputs for these metals. The findings highlight serious ecological threats and potential public health hazards due to the bioaccumulative nature of these metals. Urgent remediation, policy enforcement, and public health interventions are recommended to mitigate the risks in these vulnerable communities.

This study endeavoured to determine the current status of heavy metal contamination in soils around leather industrial areas of Tamil Nadu and the associated environmental and human health risks. Accordingly, 31 soil samples were examined using the total reflection X-ray fluorescence (TXRF) spectroscopy technique. From TXRF results, the mean concentrations of heavy metals (mg kg−1, dry weight) were obtained in the trend of Fe (9234.03) > Al (592.00) > Cr (194.99) > Sr (148.76) > Mn (137.61) > Cl (81.79) > Zn (38.97) > Cu (38.33) > V (36.78) > Pb (31.18) > Ni (9.81) > As (1.12). The pollution indices such as geo-accumulation index (Igeo), contamination factor (CF), and enrichment factor (EF) revealed the significant contaminants in the study area. The hazard index (HI) indicated that non-carcinogenic risk (NCR) due to heavy metals were negligible for both adults and children (HI < 1). In addition, total cancer risk (TCR) showed carcinogenic risk (CR) for adults (TCR = 1.38 × 10−4) and children (TCR = 1.27 × 10−3) are moderate and higher, respectively. Multivariate statistical analyses such as Pearson correlation analysis (CA), hierarchical cluster analysis (HCA), and factor analysis (FA) revealed that the heavy metals derived from various sources such as lithogenic and anthropogenic. Overall, special concern is advised to the inhabitants, especially with regard to children, in the study area about the health risks due to the anthropogenic inputs.

Characterisation of soils and establishment of heavy metal and radionuclide concentrations prior to mining disturbance provides an essential benchmark for post-mining reclamation. Land disturbance from uranium mining could result in metal contamination of surface soils with subsequent negative impacts on soil quality and human and ecosystem health. This soil characterization study was conducted because there is no background data of soil trace metals and radionuclide for this region. These benchmark data are essential for post-mining monitoring and possible future remediation. The planned conventional open pit mine will produce 15,000 and 50,000 tonnes (ore and waste rock) per day. The envisaged overburden and waste dumps are likely to enhance potential for heavy metal contamination of soils. To establish the baseline data and pre-mining site environmental conditions, (1) the physico-chemical properties of soils and (2) potential heavy metal and radionuclide pollutants in identified different sites were determined. The sampling sites were selected based on the variation in soil classes in the study area. The physico-chemical properties were measured using the standard methods while heavy metals were analysed using a Finnigan MAT Element 2 high resolution Inductively Coupled Plasma Mass Spectrometer. The following heavy metals and radionuclides were determined: Th, U, Sr, Ba, Be, Cd, Pb, V, Cr, Ni, Cu, Zn and Co. Most soils are predominately sandy clay (>17.2 %) suggesting their suitability for agricultural use. The results also indicate that most heavy metals were at or below expected ranges for non-contaminated soils and there was no evidence of any geochemical anomalies.

Heavy metals contamination in agricultural soil is a potential environmental threat to the safety of agricultural food crops such as rice which is consumed by majority of Tanzanians. The aim of this review is to put together available information on sources of heavy metals, their extent of contamination in agricultural soil and in rice, the risk of exposure through rice consumption as well as the relationship between heavy metals contamination in agricultural soils and in rice in Tanzania. There are several methods of determining the concentrations of heavy metals in soils and in rice. These include Inductively Coupled Plasma Mass Spectrometer (ICP – MS) and Energy Dispersive X – ray Fluorescence spectrometer (EDXRF). It has been mostly reported that the extent of heavy metals contamination in agricultural soils is influenced by their closeness to mining or industrial areas. The use of river waters in mining areas or wastewater from industries for irrigation has been associated to increasing levels of heavy metals in agricultural soils. The elevated level of heavy metals in agricultural soils leads to their accumulation in crops especially rice which upon consumption poses health effects to human and the ecosystem at large. This review suggests the need for determining the extent of heavy metals contamination in agricultural soils around potential areas such as mining and to link this with exposure assessment on heavy metals through rice consumption in Tanzania. This information is necessary to establish the extent at which rice consumers in Tanzania are at risk of heavy metals exposure.

Bacterial community shaped by heavy metals and contributing to health risks in cornfields

The Niger Delta environment is under serious threat due to heavy metal pollution. Many studies have been conducted on the heavy metal contamination in soils, water, seafood and plants in the Niger Delta ecosystem. However, there is a lack of clear understanding of the health consequences for people and strategies for attaining One Health, and a dispersion of information that is accessible. The study focused on investigating the contamination levels, distributions, risks, sources and impacts of heavy metals in selected regions of the Niger Delta. Prior studies revealed that the levels of certain heavy metals, including Cd, Pb, Cu, Cr, Mn, Fe and Ni, in water, sediment, fish and plants in most Niger Delta ecosystems were higher than the acceptable threshold attributed to various anthropogenic stressors. In the reviewed Niger Delta states, ecosystems in Rivers state showed the highest concentrations of heavy metals in most sampled sites. Groundwater quality was recorded at concentrations higher than 0.3mg/L World Health Organization drinking water guideline. High concentrations of copper (147.915mg/L) and zinc (10.878mg/L) were found in Rivers State. The heavy metals concentrations were greater in bottom-dwelling organisms such as bivalves, gastropods and shrimp than in other fishery species. Heavy metal exposure in the region poses risks of communicable and non-communicable diseases. Diverse remediation methods are crucial to reduce contamination levels, but comprehensive strategies and international cooperation are essential to address the health hazards. Actively reducing heavy metals in the environment can achieve One Health objectives and mitigate disease and economic burdens.

Coal consumption increases consistently, and heavy metal contamination in soils surrounding coal mines has been a severe environmental issue. Investigation of the relations of the bacterial community in soils of coal mines with potential ecological risk from heavy metals is essential for better understanding the impact of coal production on ecosystem health. In this study, soils surrounding two coal mines in a typical coal industry city of China were sampled, and heavy metals (Cd, Cr, Ni, Cu, Zn, As, and Pb) were quantified together with an in-depth evaluation of bacterial community based on high-throughput sequencing. The individual and overall ecological risk from heavy metals were estimated, and their relations with the identified seven core bacteria were analyzed. For soils surrounding both coal mines, in phylum level, five dominant bacterial taxa (Proteobacteria, Acidobacteriota, Actinobacteria, Firmicutes, and Bacteroidota) were identified, and their differences in relative abundance were clearly revealed, even if the differences between the two mines were not significant. When evaluated in genus level, apparent differences in bacterial community structure and complexity were found. The soils with higher potential ecological risk showed relatively lower complexity for the bacterial community. Seven core bacteria were identified, and negative relations of their relative abundance with the individual and overall ecological risk from heavy metals were revealed. Two families (Halomonadaceae and Aeromonadaceae) and two genera (Oceanisphaera and Ignatzschineria) were found more sensitive to both the individual risk from arsenic and the overall risk from all heavy metals, suggesting their possibility as indicators of soil contamination by these metal elements. Different from these, Rhizobiaceae revealed a positive relation with the overall risk value, implying their strong tolerance to heavy metals and, therefore, possible effect in bioremediation and phytoremediation of heavy metal contaminated soils. These findings have great reference value for better understanding the impacts of heavy metal contamination induced by coal production.

Fertilization is a vital approach to increase the crop yield by enhancing soil fertility, but some of the fertilizer sources such as pig manure contain non-essential toxic heavy metals, which can produce the environmental and public health risk. Therefore, the purpose of this study was to investigate the soil fertility and heavy metal pollution risk under long-term fertilization in acidic paddy soil. Fertilizer treatments that were arranged in randomized complete block design (RCBD) included CK (no fertilization), NK (inorganic nitrogen and potassium fertilization), NPK (inorganic NK and phosphorus fertilization), NPKM1 (70% of NPK and 30% pig manure application), NPKM2 (50% of NPK and 50% of pig manure application), and NPKM3 (30% of NPK and 70% of pig manure application). The rice grain yield and soil nutrient contents were highest under NPKM3 treatment. Long-term addition of manure significantly (P ≤ 0.05) increased soil pH and SOC content compared to the NPK fertilization. Soil available and total Cr, Cd, and Hg contents were highest under NPKM3 treatments, while soil total and available Pb content was significantly (P ≤ 0.05) higher under NPK treatment. Highest ecological risk (IR) was (1904) under NPK treatment and highest pollution load index (PLI) was 1.5 under NPKM3 treatment. Cd concentration in rice grain exceeded the maximum permissible limit of 0.1 mg kg−1 under combined application of manure and inorganic fertilization treatments. Grain Cr, Hg, and Pb contents were within safe limits of their concentration in all treatments. Moreover, biological accumulation coefficients of Cr, Cd, Hg, Pb, Zn, and Cu were highest under NPK treatment. Redundancy analysis (RDA) showed that soil pH and nutrient contents showed significant correlation with heavy metal concentrations in soil. Soil pH showed significant (P ≤ 0.05) positive effect on Cd accumulation in rice grain. Heavy metals in pig manure should be monitored before application to the field to reduce the risk of heavy metal pollution in soil and plant. Furthermore, combined application at the rate of 70% inorganic fertilization and 30% of manure could be better strategy to produce high crop yield with minimum risk of heavy metal contamination in soil and food crops.

The coastal zone of Bangladesh, with a population density of 1278 people per square kilometer, is under serious threat due to heavy metal pollution. To date, many studies have been conducted on the heavy metal contamination in soils, water, aquatic animals, and plants in the coastal zone of Bangladesh; however, the available information is dispersed. In this study, previous findings on the contamination levels, distributions, risks, and sources of heavy metals in sediments and organisms were summarized for the first time to present the overall status of heavy metal pollution along coastal regions. Earlier research found that the concentrations of various heavy metals(HMs), particularly Co, Cd, Pb, Cu, Cr, Mn, Fe, and Ni in water, sediment, and fish in most coastal locations, were above their permissible limits. High concentrations of HMs were observed in sediments and water, like Cr of 55mg/kg and 86.93mg/l in the ship-breaking areas and Karnaphuli River, respectively, in coastal regions of Bangladesh. Heavy metals severely contaminated the Karnaphuli River estuary and ship-breaking area on the Sitakundu coast, where sediments were the ultimate sink of high concentrations of metals. Sedentary or bottom-dwelling organisms like gastropods and shrimp had higher levels of heavy metals than other organisms. As a result, the modified PRISMA review method was used to look at the critical research gap about heavy metal pollution in Bangladesh's coastal areas by analyzing the current research trends and bottlenecks.

Introduction. Currently, as a result of military actions, certain regions of Ukraine face significant environmental pressures.Military activities have led to soil destruction and contamination through the construction of trenches, bunkers, and other fortifications, as well as the movement of machinery and detonation of ammunition. The contamination of soils with heavy metals and other pollutants, resulting from explosions and detonation processes, poses significant risks to plants, animals, and humans.The goal of the work. Is to explore effective methods for monitoring and mitigating technogenic impacts on soils and restoring their fertility.Materials and methods of research. Heavy metal contamination was assessed using two methods: atomic absorption spectrometry of soil samples (traditional) and analysis of heavy metals in the pollen of Melilotus albus (white sweet clover) collected by bees (proposed). White sweet clover was sown at 25 kg/ha using drones in early spring. Bee colonies were placed in safe zones, and pollen (obnizhzhia) was collected using traps, dried at 41°C, and analyzed for lead, cadmium, and zinc via atomic absorption spectrometry.Results of the research. The accumulation coefficients of heavy metals in bee pollen from Melilotus albus were: lead – 0.052, cadmium – 0.058, and zinc – 0.12. Cadmium showed the lowest accumulation rate, 8.9 times lower than lead and 2.0 times lower than zinc. Comparing methods, heavy metal levels in soil determined via bee pollen were closely aligned with direct soil analysis: lead differed by 2.7%, cadmium by 8.9%, and zinc by 9.0%.Discussion of the results. The study demonstrates the effectiveness of using Melilotus albus pollen as a bioindicator for monitoring heavy metal contamination in soils affected by technogenic factors. This approach not only provides a reliable means of assessing soil contamination but also promotes sustainable environmental management practices in post-conflict areas.Conclusion. The proposed method of soil contamination assessment, based on heavy metal content in bee pollen, offers an effective and accurate way to monitor lead, cadmium, and zinc levels. It is especially useful in high-risk areas, such as those with landmines or restricted access.

ObjectiveThis study aims to develop and compare different models to predict the Length of Stay (LoS) and the Prolonged Length of Stay (PLoS) of inpatients admitted through the emergency department (ED) in general patient settings. This aim is not only to promote any specific model but rather to suggest a decision-supporting tool (i.e., a prediction framework).MethodsWe analyzed a dataset of patients admitted through the ED to the “Sant”Orsola Malpighi University Hospital of Bologna, Italy, between January 1 and October 26, 2022. PLoS was defined as any hospitalization with LoS longer than 6 days. We deployed six classification algorithms for predicting PLoS: Random Forest (RF), Support Vector Machines (SVM), Gradient Boosting (GB), AdaBoost, K-Nearest Neighbors (KNN), and logistic regression (LoR). We evaluated the performance of these models with the Brier score, the area under the ROC curve (AUC), accuracy, sensitivity (recall), specificity, precision, and F1-score. We further developed eight regression models for LoS prediction: Linear Regression (LR), including the penalized linear models Least Absolute Shrinkage and Selection Operator (LASSO), Ridge and Elastic-net regression, Support vector regression, RF regression, KNN, and eXtreme Gradient Boosting (XGBoost) regression. The model performances were measured by their mean square error, mean absolute error, and mean relative error. The dataset was randomly split into a training set (70%) and a validation set (30%).ResultsA total of 12,858 eligible patients were included in our study, of whom 60.88% had a PloS. The GB classifier best predicted PloS (accuracy 75%, AUC 75.4%, Brier score 0.181), followed by LoR classifier (accuracy 75%, AUC 75.2%, Brier score 0.182). These models also showed to be adequately calibrated. Ridge and XGBoost regressions best predicted LoS, with the smallest total prediction error. The overall prediction error is between 6 and 7 days, meaning there is a 6–7 day mean difference between actual and predicted LoS.ConclusionOur results demonstrate the potential of machine learning-based methods to predict LoS and provide valuable insights into the risks behind prolonged hospitalizations. In addition to physicians' clinical expertise, the results of these models can be utilized as input to make informed decisions, such as predicting hospitalizations and enhancing the overall performance of a public healthcare system.