A novel prediction approach driven by graph representation learning for heavy metal concentrations.

The potential risk of heavy metals (HMs) to public health is an issue of great concern. Early prediction is an effective means to reduce the accumulation of HMs. The current prediction methods rarely take internal correlations between environmental factors into consideration, which negatively affects the accuracy of the prediction model and the interpretability of intrinsic mechanisms. Graph representation learning (GraRL) can simultaneously learn the attribute relationships between environmental factors and graph structural information. Herein, we developed the GraRL-HM method to predict the HM concentrations in soil-rice systems. The method consists of two modules, which are PeTPG and GCN-HM. In PeTPG, a graphic structure was generated using graph representation and communitization technology to explore the correlations and transmission paths of different environmental factors. Subsequently, the GCN-HM model based on the graph convolutional neural network (GCN) was used to predict the HM concentrations. The GraRL-HM method was validated by 2295 sets of data covering 21 environmental factors. The results indicated that the PeTPG model simplified correlation paths between factor nodes from 396 to 184, reducing by 53.5 % graph scale by eliminating the invalid paths. The concise and efficient graph structure enhanced the learning efficiency and representation accuracy of downstream prediction models. The GCN-HM model was superior to the four benchmark models in predicting the HM concentration in the crop, improving R2 by 36.1 %. This study develops a novel approach to improve the prediction accuracy of pollutant accumulation and provides valuable insights into intelligent regulation and planting guidance for heavy metal pollution control.

Ensemble learning-based applied research on heavy metals prediction in a soil-rice system.

Accurate prediction of heavy metal accumulation in soil ecosystems is crucial for maintaining healthy soil environments and ensuring high-quality agricultural products, as well as a challenging scientific task. In this study, we constructed a dataset containing 490 sets of multidimensional environmental covariate data and proposed prediction models for heavy metal concentrations (HMC) in a soil-rice system, EL-HMC (including RF-HMC and GBM-HMC), based on Random Forest (RF) and Gradient Boosting Machine (GBM) ensemble learning (EL) techniques. To reasonably evaluate the effectiveness of each model, Multiple linear and Bayesian regressions were selected as benchmark models (BM), and mean absolute error (MAE), root mean square error (RMSE), and determination coefficient R2 were selected as evaluation indicators. In addition, sensitivity and spatial autocorrelation (SAC) analyses were used to examine the robustness of the model. The results showed that the R2 values of RF-HMC and GBM-HMC for modeling available cadmium (Cd) concentrations in soil were 0.654 and 0.690, respectively, with an average increase of 48.0 % compared to the BMs. The R2 values of RF-HMC and GBM-HMC for predicting Cd, lead (Pb), chromium (Cr), and mercury (Hg) concentrations in rice ranged from 0.618 to 0.824 and 0.645 to 0.850, respectively, with an average increase of 58.2 % compared with the BMs. The corresponding MAEs and RMSEs of RF-HMC and GBM-HMC had low error levels. Sensitivity analysis of the input features and the SAC of the prediction bias showed that the EL-HMC models have excellent robustness. Therefore, the EL technology-based prediction models for HMCs proposed herein are practical and feasible, demonstrating better accuracy and stability than the traditional model. This study verifies the application potential of EL technology in pollution ecology and provides a new perspective and solution for sustainable management and precise prevention of heavy metal pollution in farmland soil at the regional scale.

Driving effects and transfer prediction of heavy metal(loid)s in contaminated courtyard gardens using redundancy analysis and multilayer perceptron.

The distribution and migration of heavy metal(loid)s in the soil-vegetable systems of courtyard gardens near mining areas have rarely been investigated, leading to potential food safety risks for residents. Moreover, the existing research is mainly focused on the total content of heavy metal(loid)s (tMetals) rather than the bioavailable contents (aMetals). In this study, 26 and 28 pairs of soil and vegetable samples were collected from the courtyard gardens near the Realgar mine in Baiyun Town and the lead-zinc (Pb-Zn) mine in Shuikoushan Town, respectively. The tMetal and aMetal of cadmium (Cd), mercury (Hg), arsenic (As), Pb, chromium (Cr), nickel (Ni), copper (Cu), Zn, manganese (Mn), iron (Fe), and calcium (Ca) in the samples were analyzed in this study. The results showed that courtyard gardens were polluted by various heavy metal(loid)s at varying degrees. The bioavailabilities of different metals varied significantly, among which Cd has the highest bioavailability (> 30%). In the transfer process of heavy metal(loid)s, the transfer rate (Tf) was ranked as soil-roots (1.50) > stems-leaves (1.07) > roots-stems (0.46) > stems-fruits (0.33). Redundancy analysis was used to evaluate the driving effects, and the results revealed that aCa, aZn, and aFe in soil could inhibit the absorption of aCd by plant roots. Soil organic matter was the inhibiting factor regarding the transfer of aAs and aCu, whereas it was also the promoting factor for transferring aPb, aNi, and aCr. Furthermore, the multilayer perceptron (MLP) could effectively predict the Tf of heavy metal(loid)s based on the aMetal. The R2 values of the MLP were ranked as follows: 0.91 for As, 0.88 for Zn, 0.85 for Hg, 0.83 for Cu, 0.79 for Cr, 0.66 for Cd, 0.65 for Pb, and 0.52 for Ni. This study emphasizes the aMetal-based ecological characteristics and prediction ability. The study results are significant for guiding residents to strategize appropriate crop planting and ensure the safe production and consumption of vegetables.

Convolutional neural networks-based health risk modelling of some heavy metals in a soil-rice system.

The long-term consumption of heavy metal-rich rice can cause serious harm to human health. However, the existing health risk assessment (HRA) can only be performed after the rice has been harvested, and this approach belongs to a passive and lagging pattern. This study is the first to explore the feasibility of health risk (HR) prediction by proposing the indirect model CNNHR-IND and the direct model CNNHR-DIR based on the convolutional neural network (CNN) technology. The dataset included 390 pairs of soil-rice samples collected from You County, China, with 17 environmental covariates. The R2 values for CNNHR-IND for non-carcinogenic and carcinogenic risks were 0.578 and 0.554, respectively, and those for CNNHR-DIR were 0.647 and 0.574, respectively. The results demonstrated that both models performed well, especially CNNHR-DIR had a higher estimation accuracy. The spatial autocorrelation analysis indicated that CNNHR-DIR exerted no systematic bias in the prediction results for health risks, confirming the rationality of the CNNHR-DIR model. The sensitivity analysis further confirmed the generalizability and robustness of CNNHR-DIR. This study proved the feasibility of HR prediction and the potential of CNN technology in HRA, and is significant regarding early risk warnings of rice planting and the sustainable development of public health.

A field study to estimate heavy metal concentrations in a soil-rice system: Application of graph neural networks.

Accurate prediction of the concentration of heavy metals is of great significance for assessing the quality of agricultural products and reducing health risks. However, the complexity and interconnectivity of the farmland ecosystem restricts the improvement of the prediction accuracy of traditional methods. This research explored the application potential of graph neural network (GNN) technology, which can extract and learn information in large-scale networks in detail, in the field of heavy metal prediction for the first time. In this study, a heavy metal prediction model for rice, CoNet-GNN, was proposed with 17 environmental factors as input variables using the co-occurrence network and GNN. Experimental results using a dataset from a field study showed that the R2 of CoNet-GNN for predicting Cd, Pb, Cr, As, and Hg had outstanding values of 0.872, 0.711, 0.683, 0.489, and 0.824, respectively. Sensitivity analysis further indicated that CoNet-GNN had good stability and robustness. Compared with random forest, gradient boosting, and multilayer perceptron, CoNet-GNN made a remarkable improvement to the prediction accuracy of all studied heavy metals. Therefore, CoNet-GNN can effectively simulate the rich relationships and laws between various factors in the soil-rice system and effectively characterize the influence diffusion path. Furthermore, it provides new ideas for heavy metal prediction based on network research methods and expands the technical scope of heavy metal evaluation.

Probabilistic health risk assessment for residents exposed to potentially toxic elements near typical mining areas in China.

Public health problems caused by toxic elements in mining areas have always been an important topic worldwide. However, existing studies have focused on single exposure routes and common toxic elements, which might underestimate the risks faced by residents. In this study, three typical mining areas in central China were selected to assess the health risks of 14 potentially toxic elements through five exposure routes using Monte Carlo simulations. The results indicated that the 95th percentile non-carcinogenic risk values to humans via rice and vegetable ingestion ranged from 9.8 to 26.0 and 6.2 to 19.0. The corresponding carcinogenic risks ranged from 1.4E-2 to 6.3E-2 and from 2.9E-3 to 2.3E-2, respectively. Therefore, residents face serious health risks. Multi-element analysis showed that cadmium (Cd), boron (B), and arsenic (As) were the main contributors to rice non-carcinogenicity, whereas Cd and nickel (Ni) were the main elements of rice carcinogenicity. B and lead (Pb) played an essential role in the non-carcinogenesis of vegetables, and B, Ni, and Cd played an essential role in carcinogenesis. Accidental ingestion is the main route of soil exposure. In these three areas, the probability of non-carcinogenic risk faced by adults was 40%, 0%, and 1%, respectively, while the probabilities for children were 100%, 62%, and 83%, respectively. Regarding carcinogenicity, the risk for both adults and children was up to 100%. This study emphasizes the overall health risks in polluted areas via multi-route and multi-element analysis. This conclusion is helpful to comprehensively assess the potential health risks faced by residents in mining areas and provide baseline data support and a scientific basis for formulating reasonable risk control measures.

Convolutional neural network-based applied research on the enrichment of heavy metals in the soil-rice system in China.

The enrichment of heavy metals in the soil-rice system is affected by various factors, which hampers the prediction of heavy metal concentrations. In this research, a prediction model (CNN-HM) of heavy metal concentrations in rice was constructed based on convolutional neural network (CNN) technology and 17 environmental factors. For comparison, other machine learning models, such as multiple linear regression, Bayesian ridge regression, support vector machine, and backpropagation neural networks, were applied. Furthermore, the LH-OAT method was used to evaluate the sensitivity of CNN-HM to each environmental factor. The results showed that the R2 values of CNN-HM for Cd, Pb, Cr, As, and Hg were 0.818, 0.709, 0.688, 0.462, and 0.816, respectively, and both the MAE and RMAE values were acceptable. The sensitivity analysis showed that the concentrations of Cd and Pb, mechanical composition, soil pH, and altitude were the main sensitive features for CNN-HM. Compared with CNN-HM based on all input features, the performance of the quick prediction model that was based on the sensitive features did not degrade significantly, thereby indicating that CNN-HM has stronger stability and robustness. The quick prediction model has extensive application value for timely prediction of the enrichment of heavy metals in emergencies. This study demonstrated the effectiveness and practicability of CNNs in predicting heavy metal enrichment in the soil-rice system and provided a new perspective and solution for heavy metal prediction.

Probabilistic health risk assessment of inorganic arsenic and some heavy metals in rice produced from a typical multi-mining county, China.

The potential impact of exposure to toxic elements in rice on human health has become a global public health issue. This study analyzed the pollution characteristics and probabilistic health risks of exposure to iAs, Pb, Cd, Cr, and Hg in rice produced in a typical multi-mining county using Monte Carlo simulation, a geographic information system, and bioavailability analysis. The results showed that the enrichment of As and Cd was prominent in rice, with mean tAs, iAs, and Cd contents of 0.34 ± 0.20, 0.15 ± 0.09, and 0.48 ± 0.50 mg·kg-1, respectively. The probability of non-carcinogenic risk via rice consumption in adults and children exceeding the threshold was 72% and 78%, respectively, and that of carcinogenic risk was as high as 100%. Among toxic elements, Cd and iAs were the main risk factors for health risks. The high-level health-risk areas mainly occurred in the north-eastern and central parts of the study area. Sensitivity analysis highlighted that the top three influential parameters for non-carcinogenic risk in adults were Content(Cd), Content(iAs), and Bioaccessibility(Cd), whereas those in children were ingestion rate of rice, Content(Cd), and Content(iAs). The Content(Cd) was the decisive factor for carcinogenic risk, with a sensitivity coefficient of 78.0% in adults and 64.7% in children. Considering the high risk of ingestion of local rice in this area, it is imperative to place strict supervision and take control measures.