Prediction of sulfate concentrations in groundwater in areas with complex hydrogeological conditions based on machine learning

The persistent and increasing levels of sulfate due to a variety of human activities over the last decades present a widely concerning environmental issue. Understanding the controlling factors of groundwater sulfate and predicting sulfate concentration is critical for governments or managers to provide information on groundwater protection. In this study, the integration of self-organizing map (SOM) approach and machine learning (ML) modeling offers the potential to determine the factors and predict sulfate concentrations in the Huaibei Plain, where groundwater is enriched with sulfate and the areas have complex hydrogeological conditions. The SOM calculation was used to illustrate groundwater hydrochemistry and analyze the correlations among the hydrochemical parameters. Three ML algorithms including random forest (RF), support vector machine (SVM), and back propagation neural network (BPNN) were adopted to predict sulfate levels in groundwater by using 501 groundwater samples and 8 predictor variables. The prediction performance was evaluated through statistical metrics (R2, MSE and MAE). Mine drainage mainly facilitated increase in groundwater SO42- while gypsum dissolution and pyrite oxidation were found another two potential sources. The major water chemistry type was Ca-HCO3. The dominant cation was Na+ while the dominant anion was HCO3-. There was an intuitive correlation between groundwater sulfate and total dissolved solids (TDS), Cl-, and Na+. By using input variables identified by the SOM method, the evaluation results of ML algorithms showed that the R2, MSE and MAE of RF, SVM, BPNN were 0.43-0.70, 0.16-0.49 and 0.25-0.44. Overall, BPNN showed the best prediction performance and had higher R2 values and lower error indices. TDS and Na+ had a high contribution to the prediction accuracy. These findings are crucial for developing groundwater protection and remediation policies, enabling more sustainable management.

Implication of self-organizing map, stable isotopes combined with MixSIAR model for accurate nitrogen control in a well-protected reservoir

Nitrogen pollution and eutrophication in reservoirs is a global environmental geochemical concern. Occasional algal blooms still exist in reservoirs that have undergone pollution treatment. The lack of quantitative evidence of nitrogen sources and fate limits long-term stable ecological safety management. This work applied an approach integrated zonal mapping, stable isotopes (δ18OH2O, δ15Nnitrate, δ18Onitrate, and δ13C-DIC) and a Bayesian isotope model to analyze regional and seasonal differences in the contribution and sources of nitrogen to a well-protected reservoir. The values of δ18Onitrate and the positive relationship between NO3- and δ13C-DIC suggested that nitrification was the primary NO3- production in the rivers. While Denitrification was present at only a few sites. Results of the MixSIAR model coupled the NO3-/Cl- indicator revealed that the domestic sewage contributed high riverine NO3- loading (68.6 ± 10.6 %) in the dry season. In the wet season, the main nitrate sources of upper watershed were ammonia and carbamide fertilizers (47.5 % and 40.3 %). While the domestic sewage was still the major contributor of downstream region (a dense residential area), indicating possible problems with rainwater and sewage drainage networks. The results implied that the colleting and treatment of sewages were the priority in downstream region, and non-point source pollution control and wastewater treatment plant upgrading were essential to control nitrate pollution in the two upstream regions. These findings provide new insights into precise nitrogen pollution traceability and identification of treatment priorities in the sub-region, and promote the management other well-protected watershed in similar need of further nitrogen contamination control.

Groundwater hydrochemistry, source identification and health assessment based on self-organizing map in an intensive mining area in Shanxi, China

The Changzhi Basin in Shanxi is renowned for its extensive mining activities. It's crucial to comprehend the spatial distribution and geochemical factors influencing its water quality to uphold water security and safeguard the ecosystem. However, the complexity inherent in hydrogeochemical data presents challenges for linear data analysis methods. This study utilizes a combined approach of self-organizing maps (SOM) and K-means clustering to investigate the hydrogeochemical sources of shallow groundwater in the Changzhi Basin and the associated human health risks. The results showed that the groundwater chemical characteristics were categorized into 48 neurons grouped into six clusters (C1-C6) representing different groundwater types with different contamination characteristics. C1, C3, and C5 represent uncontaminated or minimally contaminated groundwater (Ca-HCO3 type), while C2 signifies mixed-contaminated groundwater (HCO3-Ca type, Mixed Cl-Mg-Ca type, and CaSO4 type). C4 samples exhibit impacts from agricultural activities (Mixed Cl-Mg-Ca), and C6 reflects high Ca and NO3- groundwater. Anthropogenic activities, especially agriculture, have resulted in elevated NO3- levels in shallow groundwater. Notably, heightened non-carcinogenic risks linked to NO3-, Pb, F-, and Mn exposure through drinking water, particularly impacting children, warrant significant attention. This research contributes valuable insights into sustainable groundwater resource development, pollution mitigation strategies, and effective ecosystem protection within intensive mining regions like the Changzhi Basin. It serves as a vital reference for similar areas worldwide, offering guidance for groundwater management, pollution prevention, and control.

Transcriptomic profiling of sciatic nerves and dorsal root ganglia reveals site-specific effects of prediabetic neuropathy

Peripheral neuropathy (PN) is a severe and frequent complication of obesity, prediabetes, and type 2 diabetes characterized by progressive distal-to-proximal peripheral nerve degeneration. However, a comprehensive understanding of the mechanisms underlying PN, and whether these mechanisms change during PN progression, is currently lacking. Here, gene expression data were obtained from distal (sciatic nerve; SCN) and proximal (dorsal root ganglia; DRG) injury sites of a high-fat diet (HFD)-induced mouse model of obesity/prediabetes at early and late disease stages. Self-organizing map and differentially expressed gene analyses followed by pathway enrichment analysis identified genes and pathways altered across disease stage and injury site. Pathways related to immune response, inflammation, and glucose and lipid metabolism were consistently dysregulated with HFD-induced PN, irrespective of injury site. However, regulation of oxidative stress was unique to the SCN while dysregulated Hippo and Notch signaling were only observed in the DRG. The role of the immune system and inflammation in disease progression was supported by an increase in the percentage of immune cells in the SCN with PN progression. Finally, when comparing these data to transcriptomic signatures from human patients with PN, we observed conserved pathways related to metabolic dysregulation across species, highlighting the translational relevance of our mouse data. Our findings demonstrate that PN is associated with distinct site-specific molecular re-programming in the peripheral nervous system, identifying novel, clinically relevant therapeutic targets.

Using unsupervised machine learning models to drive groundwater chemistry and associated health risks in Indo-Bangla Sundarban region

Groundwater is an essential resource in the Sundarban regions of India and Bangladesh, but its quality is deteriorating due to anthropogenic impacts. However, the integrated factors affecting groundwater chemistry, source distribution, and health risk are poorly understood along the Indo-Bangla coastal border. The goal of this study is to assess groundwater chemistry, associated driving factors, source contributions, and potential non-carcinogenic health risks (PN-CHR) using unsupervised machine learning models such as a self-organizing map (SOM), positive matrix factorization (PMF), ion ratios, and Monte Carlo simulation. For the Sundarban part of Bangladesh, the SOM clustering approach yielded six clusters, while it yielded five for the Indian Sundarbans. The SOM results showed high correlations among Ca2+, Mg2+, and K+, indicating a common origin. In the Bangladesh Sundarbans, mixed water predominated in all clusters except for cluster 3, whereas in the Indian Sundarbans, Cl--Na+ and mixed water dominated in clusters 1 and 2, and both water types dominated the remaining clusters. Coupling of SOM, PMF, and ionic ratios identified rock weathering as a driving factor for groundwater chemistry. Clusters 1 and 3 were found to be influenced by mineral dissolution and geogenic inputs (overall contribution of 47.7%), while agricultural and industrial effluents dominated clusters 4 and 5 (contribution of 52.7%) in the Bangladesh Sundarbans. Industrial effluents and agricultural activities were associated with clusters 3, 4, and 5 (contributions of 29.5% and 25.4%, respectively) and geogenic sources (contributions of 23 and 22.1% in clusters 1 and 2) in Indian Sundarbans. The probabilistic health risk assessment showed that NO3- poses a higher PN-CHR risk to human health than F- and As, and that potential risk to children is more evident in the Bangladesh Sundarban area than in the Indian Sundarbans. Local authorities must take urgent action to control NO3- emissions in the Indo-Bangla Sundarbans region.

Quantification of toxic metals in chicken egg and chicken feed via SOM-artificial neural network

Poultry products such as meat and eggs are rich sources of proteins, vitamins, and minerals. It is a good indicator of healthy food. Keeping in view, the present study is designed to evaluate the prevalence of toxic heavy metals (lead, nickel, cadmium, and chromium) in chicken eggs and feed. For this purpose, five samples of egg and feed were collected from five different commercial markets in Skardu City. Each sample was prepared using the wet digestion method and analyzed using a flame atomic absorption spectrophotometer. The results showed that lead, nickel, and chromium were present in varying amounts in the feed and egg, with nickel being the most concentrated metal, followed by lead and chromium in egg samples, while the feed samples showed the highest concentration of chromium followed by lead and nickel. However, concentrations of selected heavy metals except cadmium were all above the permissible limit of the World Health Organization. The self-organizing map-artificial neural network is employed for the identification of patterns of heavy metals in chicken feed and egg samples. The lower left neurons of the maps showed higher heavy metal concentrations found in samples taken from Bazar, whereas the rest of the samples showed varied concentrations. A comparison of feed and egg concentrations showed that nickel concentration was lower in feed samples than in egg samples. The lead concentration decreased in eggs except in the Krasmathang feed sample. Chromium concentration presented a negative correlation due to the extremely high concentration found in the Bazar feed sample.

Analysis of self-organizing maps and explainable artificial intelligence to identify hydrochemical factors that drive drinking water quality in Haor region

Water contamination undermines human survival and economic growth. Water resource protection and management require knowledge of water hydrochemistry and drinking water quality characteristics, mechanisms, and factors. Self-organizing maps (SOM) have been developed using quantization and topographic error approaches to cluster hydrochemistry datasets. The Piper diagram, saturation index (SI), and cation exchange method were used to determine the driving mechanism of hydrochemistry in both surface and groundwater, while the Gibbs diagram was used for surface water. In addition, redundancy analysis (RDA) and a generalized linear model (GLM) were used to determine the key drinking water quality parameters in the study area. Additionally, the study aimed to utilize Explainable Artificial Intelligence (XAI) techniques to gain insights into the relative importance and impact of different parameters on the entropy water quality index (EWQI). The SOM results showed that thirty neurons generated the hydrochemical properties of water and were organized into four clusters. The Piper diagram showed that the primary hydrochemical facies were HCO3--Ca2+ (cluster 4), Cl---Na+ (all clusters), and mixed (clusters 1 and 4). Results from SI and cation exchange show that demineralization and ion exchange are the driving mechanisms of water hydrochemistry. About 45 % of the studied samples are classified as "medium quality"," that could be suitable as drinking water with further refinement. Cl- may pose increased non-carcinogenic risk to adults, with children at double risk. Cluster 4 water is low-risk, supporting EWQI findings. The RDA and GLM observations agree in that Ca2+, Mg2+, Na+, Cl- and HCO3- all have a positive and significant effect on EWQI, with the exception of K+. TDS, EC, Na+, and Ca2+ have been identified as influencing factors based on bagging-based XAI analysis at global and local levels. The analysis also addressed the importance of SO4, HCO3, Cl, Mg2+, K+, and pH at specific locations.

Source apportionment and migration characteristics of heavy metal(loid)s in soil and groundwater of contaminated site

The rapid industrial growth has generated heavy metal(loid)s contamination in the soil, which poses a serious threat to the ecology and human health. In this study, 580 samples were collected in Henan Province, China, for source apportionment, migration characterization and health risk evaluation using self-organizing map, positive matrix factorization and multivariate risk assessment methods. The results showed that samples were classified into four groups and pollution sources included chromium slag dump, soil parent rock and abandoned factory. The contents of Cr, Pb, As and Hg were low in Group 1. Group 2 was characterized by total Cr, Cr(Ⅵ) and pH. The enrichment of total Cr and Cr(Ⅵ) in soil was mainly attributed to chromium slag dump, accounting for more than 84.0%. Group 3 was dominated by Hg and Pb. Hg and Pb were primarily attributed to abandoned factory, accounting for 84.7% and 70.0%, respectively. Group 4 was characterized by As. The occurrence of As was not limited to one individual region. The contribution of soil parent rock reached 83.0%. Furthermore, the vertical migration of As, Hg, Pb and Cr(Ⅵ) in soil was mainly influenced by medium permeability, pH and organic matter content. The trends of As, Pb, and Hg with depth were basically consistent with the trends of organic matter with depth, and were negatively correlated with the change in pH with depth. The trends of Cr(Ⅵ) with depth were basically consistent with the changes in pH with the depth. The content of Cr(Ⅵ) in the deep soil did not exceed the detection limits and Cr(Ⅵ) contamination occurred in the deep aquifer, suggesting that Cr(Ⅵ) in the deep groundwater originated from the leakage of shallow groundwater. The assessment indicated that the non-carcinogenic and carcinogenic risks for children and adults could not be neglected. Moreover, children were more susceptible than adults.

An integrated approach for quantifying trace metal sources in surface soils of a typical farmland in the three rivers plain, China

The presence of trace metals (TMs) in agricultural soil has garnered considerable attention due to their potential migration into crops, posing a significant risk to human health. In this study, we examined the concentrations of eight trace metals (Cd, Cr, Cu, Hg, Mn, Ni, Pb, and Zn) in the soil and investigated various soil physicochemical characteristics in the Three Rivers Plain region, China. The assessment of the geoaccumulation index (Igeo) for the mean concentration of all trace metals indicated that the soils were generally free from significant TM pollution. However, a noteworthy finding emerged in relation to Hg, where the maximum Igeo value suggested moderate pollution levels. Kriging prediction results further indicated that approximately 1.55% of the study area might be impacted by Hg pollution. Moreover, it is prudent to direct attention towards Cd, Cr, Cu, Mn, and Ni, as their Igeo values revealed that the region with the highest concentrations of these metals ranged from unpolluted to moderately polluted. This study employed a comprehensive approach, utilizing the Self-Organizing Map (SOM), Kriging spatial distribution, and the Positive Matrix Factorization (PMF) model to identify the sources of TMs in agricultural soil. The results unveiled that the primary contributors to TM presence were the natural parental materials, alongside industrial activities such as coal mining and coal plant operations, as well as agricultural practices. These findings provide foundational insights for future management strategies in the Three Rivers Plain, aiming to enhance agricultural productivity and promote sustainability.

Remediation potential of magnetic biochar in lead smelting sites: Insight from the complexation of dissolved organic matter with potentially toxic elements

Magnetic biochar has been widely used in potentially toxic elements (PTEs) polluted soils due to its magnetic separation capability and synchronous immobilization for multiple metals. However, the contribution of magnetic biochar to soil dissolve organic material (SDOM) and its binding behavior with PTEs needs to be further clarified prior to its remediation application on lead smelting sites. In this study, multi-spectral techniques of excitation-emission matrix (EEM) fluorescence spectroscopy and two-dimensional FTIR correlation spectroscopy (2D-FTIR-COS) were used to explore the evolution characteristics of SDOM in the lead smelting site under the remediation of magnetic biochar, and to further analyze its affinity and binding behavior with Pb and As. Results showed that magnetic biochar significantly increased SDOM content and decreased Pb and As available content. EEM and parallel factor analysis (EEM-PARAFAC) and Self-Organizing map analysis showed that humus-like and aromatic DOM increased and microbial-derived SDOM decreased after magnetic biochar cultivation. Furthermore, 2D-FTIR-COS correlation spectroscopy analysis indicated that BDOM had a stronger binding affinity to Pb, while SDOM has a stronger binding affinity to As. The binding sequences of different DOMs to PTEs varied greatly, the carboxyl and amide groups of SDOM and BDOM showed a remarkable and rapid response. Our results enhance the insights of magnetic biochar on soil function and PTEs remediation potential, providing novel information for its environmental remediation application.

Community recovery of benthic macroinvertebrates in a stream influenced by mining activity: Importance of microhabitat monitoring

The decrease in freshwater biodiversity owing to anthropogenic disturbances such as mining activity is a global challenge; hence, there is an urgent need for systematic approaches to continuously monitor such disturbances and/or the recovery of biodiversity in freshwater habitats. The Hwangjicheon Stream is the source of South Korea's longest river and has been subjected to runoff from coal mining. We investigated changes in the diversity of the benthic macroinvertebrate community in various microhabitats, including riffle, run, and pool, to monitor the recovery of biodiversity in the stream following the improvement of a mining water treatment plant in 2019. The dataset comprised 111 samples obtained from four types of microhabitats (riffle, run, pool, and riparian) over a four-year period from 2018 to 2021. The mining-affected sites had lower macroinvertebrate community complexities according to a network analysis, and grouped into the same cluster based on self-organizing map (SOM) analysis. Moreover, 51 taxa selected as indicator species represented each cluster obtained through the SOM analysis. Among them, only Limnodrilus gotoi and Radix auricularia were included as indicator species at the mining-affected sites. However, after 2020, the benthic macroinvertebrate community complexity increased, and some of the microhabitats at the mining-affected sites were included in the same cluster as the reference sites in the SOM analysis, indicating that the recovery of benthic macroinvertebrate communities had initiated in certain microhabitats (e.g., riparian). Further analysis confirmed that the macroinvertebrate community clearly differed according to the survey year, even in different microhabitats at the same sites. This suggests that more acute microhabitat monitoring may be necessary to quickly confirm biodiversity restoration when assessing the degree of the recovery in river biodiversity from anthropogenic disturbances.

Transcriptomic analysis of diabetic kidney disease and neuropathy in mouse models of type 1 and type 2 diabetes

Diabetic kidney disease (DKD) and diabetic peripheral neuropathy (DPN) are common complications of type 1 (T1D) and type 2 (T2D) diabetes. However, the mechanisms underlying pathogenesis of these complications are unclear. In this study, we optimized a streptozotocin-induced db/+ murine model of T1D and compared it to our established db/db T2D mouse model of the same C57BLKS/J background. Glomeruli and sciatic nerve transcriptomic data from T1D and T2D mice were analyzed by self-organizing map and differential gene expression analysis. Consistent with prior literature, pathways related to immune function and inflammation were dysregulated in both complications in T1D and T2D mice. Gene-level analysis identified a high degree of concordance in shared differentially expressed genes (DEGs) in both complications and across diabetes type when using mice from the same cohort and genetic background. As we have previously shown a low concordance of shared DEGs in DPN when using mice from different cohorts and genetic backgrounds, this suggests that genetic background may influence diabetic complications. Collectively, these findings support the role of inflammation and indicate that genetic background is important in complications of both T1D and T2D.

The spatial analysis, risk assessment and source identification for mercury in a typical area with multiple pollution sources in southern China

Mercury (Hg) has always been a research hot spot because of its high toxicity. This study conducted in farmland near rare earth mining area and traffic facilities, which considered multiple pollution sources innovatively. It not only analyzed Hg spatial characteristics using inverse distance weighting and self-organizing map (SOM), but also assessed its pollution risk by potential ecological risk index (Er) as well as geoaccumulation index (Igeo), and identified the pollution sources with positive matrix factorization. The results showed that there was no heavy Hg pollution in most farmland, while a few sampling sites with Hg pollution were close to highway, railway station and petrol station in Xinfeng or in the farmland of Anyuan, which were divided into the cluster with highest Hg concentration in SOM. The vehicle exhaust emission and pesticide as well as fertilizer additions significantly contributed to the local Hg pollution. Besides, there was moderate pollution and high ecological risk in Anyuan assessed by Igeo and Er, respectively. In contrast, Xinfeng had the moderate and considerable ecological risks in a larger scale. The enriched Hg might harmed not only the nearby ecological environment, but also the human health when it entered human body through food chain. The three factors that contributed to mercury concentration in this area according to positive matrix factorization were natural source, traffic source and agricultural source, respectively. This study about Hg pollution in the typical area would provide scientific evidence for the particular treatment of Hg pollution from various pollution sources like traffic source, agricultural source, etc.

An extreme wind speed climatology - Atmospheric driver identification using neural networks

Extreme wind speeds are a significant climate risk, potentially endangering human lives, causing damage to infrastructure, affecting maritime and aviation activity, along with the optimal operation of wind energy conversion systems. In this context, accurate knowledge of return levels for various return periods of extreme wind speeds and their atmospheric circulation drivers is essential for effective risk management. In this paper, location-specific extreme wind speed thresholds are identified and return levels of extremes are estimated using the Peaks-Over-Threshold method of the Extreme Value Analysis framework. Furthermore, using an environment-to-circulation approach, the key atmospheric circulation patterns that cause extreme wind speeds are identified. The data used for this analysis are hourly wind speed data, mean sea level pressure and geopotential at 500 hPa from the ERA5 reanalysis dataset, at a horizontal resolution of 0.25° × 0.25°. The thresholds are selected utilizing the Mean Residual Life plots, while the exceedances are modeled with the General Pareto Distribution. The diagnostic metrics exhibit satisfactory goodness-of-fit and the maxima of extreme wind speed return levels are located over marine and coastal areas. The optimal Self-Organizing-Map (2 × 2) is selected using the Davies-Bouldin criterion, and the atmospheric circulation patterns are related to the cyclonic activity in the area. The proposed methodological framework can be applied to other areas, that are endangered by extreme phenomena or in need of accurately assessing the principal drivers of extremes.

Assessment of heavy metal contamination of an electrolytic manganese metal industrial estate in northern China from an integrated chemical and magnetic investigation

Heavy metal concentrations (Al, V, Mn, Fe, Co, Ni, Cu, Zn, and Pb) and the magnetic properties of soil and sediment samples in/around an electrolytic manganese metal (EMM) industrial estate in northern China were investigated. Potential enrichment of Mn, Zn, and Pb was found in/around the core area of the EMM industrial estate; however, the pollution load index (PLI) values did not indicate severely polluted levels. For adults, all hazard index (HI) values of noncarcinogenic risks in the soil samples were below the safe level of 1.00. For children, none of the HI values exceeded the safe level, except Mn (HI = 1.23) in one industrial estate sample. The particle size of magnetic materials was mostly in the range of stable single-domain, and coarser ferrimagnetic phases enhanced the magnetic parameters in the industrial estate soils. Highly positive correlations were found between magnetic parameters, heavy metal concentrations, and PLI values, demonstrating that the magnetic parameters are an efficient proxy for assessing heavy metal contamination. Enrichment of Mn, Zn, and Pb was mainly derived from the EMM industry. The data showed that the EMM industrial estate under cleaner production had limited adverse impacts on the adjacent environment from the perspective of heavy metal contamination.

Estimation of near-surface ozone concentration and analysis of main weather situation in China based on machine learning model and Himawari-8 TOAR data

Ozone (O₃) is an important greenhouse gas in the atmosphere. Stratospheric ozone protects human beings, but high near-surface ozone concentrations threaten environment and human health. Owing to the uneven distribution of ground-monitoring stations and the low time resolution of polar orbiting satellites, it is difficult to accurately evaluate the refinement and synergistic pollution of near-surface ozone in China. Besides, atmospheric circulation patterns also affect ozone concentrations greatly. In this study, a new generation of geostationary satellite is used to estimate the hourly near-surface ozone concentration with a spatial resolution of 0.05°. First, the Pearson correlation coefficient and maximum information coefficient were used to study the correlation between the top of atmospheric radiation (TOAR) of Himawari-8 satellite and O₃ concentration; seven TOAR channels were selected. Second, based on an interpretable deep learning model, the hourly ozone concentration in China from September 2015 to August 2021 was obtained using the TOAR-O₃ model. Finally, the self-organizing map method was used to determine six major summer weather circulation patterns in China. The results showed that (1) the near-surface O₃ concentration can be accurately estimated; the R² (RMSE: μg/m³) values of the daily, monthly, and annual tenfold cross validation results were 0.91 (12.74), 0.97 (5.64), and 0.98 (1.75), respectively. The feature importance of the model showed that the temperature, TOAR, and boundary layer height contributed 38 %, 22 %, and 13 %, respectively. (2) The O₃ concentration showed obvious spatiotemporal difference and gradually increased from 10:00 to 15:00 (Beijing time) every day. In most areas of China, O₃ concentration had increased significantly. (3) The O₃ concentration in northern China was the highest under the circulation pattern of the Meiyu front over the Yangtze River Delta, while in southern China, it was the highest under the circulation pattern of the northeast cold vortex controlling most of China.

Machine learning based on the graph convolutional self-organizing map method increases the accuracy of pollution source identification: A case study of trace metal(loid)s in soils of Jiangmen City, south China

Rapid economic development and industrialization may include environmentally harmful human activities that cause heavy-metal accumulation in soils, ultimately threatening the quality of the soil environment and human health. Therefore, accurate identification of pollution sources is an important weapon in efforts to control and prevent pollution. The self-organizing map (SOM) method is widely used in pollution source identification because of its capacity for visualization of high-dimensional data. The SOM ignores the graph structure relationship among chemical elements in soils; the SOM analysis of pollution sources has high uncertainty. Here, we propose a new analysis method, i.e., the graph convolutional self-organizing map (GCSOM), which uses a graph convolutional network (GCN) to extract the graph structure relationship among the chemical elements in soils, then performs data visualization using an SOM. We compared the performances of GCSOM and SOM, then assessed the pollution source characteristics of trace metal(loid)s (TMs, mostly heavy metals) in Jiangmen City using the GCSOM. Our experimental results showed that the GCSOM is superior to the SOM for identification of TM sources, while the TMs in the soil of Jiangmen originate from three main sources: agricultural activities (mainly in Taishan City, Jiangmen), traffic emissions (mainly in Xinhui and Pengjiang Districts), and industrial activities (mainly in Xinhui District). The risk assessment indicated that the risk of all TMs was within threshold.

Analysis of water quality indexes and their relationships with vegetation using self-organizing map and geographically and temporally weighted regression

Natural vegetation has been proved to promote water purification in previous studies, while the relevant laws has not been excavated systematically. This research explored the relationships between vegetation cover and water quality indexes in Liaohe River Basin in China combined with self-organizing map (SOM) and geographically and temporally weighted regression (GTWR) innovatively and systematically based on the distributing heterogeneity of water quality conditions. Results showed that the central and northeast regions of the study area had serious organic and nutrient pollution, which needed targeted treatment. And SOM verified that high vegetation coverage with retention potential of organic and inorganic pollutants as well as nutrients improved water quality to some degree, while the excessive discharges of pollutants still had serious threats to nearby water environment despite the purification function of vegetation. GTWR indicated that the waterside vegetation was beneficial for dissolved oxygen increasing and contributed to the decreasing of organic pollutants and inorganic pollutants with reducibility. Natural vegetation also obsorbed nutrients like TN and TP to some degree. However, the retential potential of nitrogen and organic pollutants became not obvious when there were heavy pollution, which demonstrated that pollution sources should be controlled despite the purification function of vegetation. This study implied that natural vegetation purified water quality to some degree, while this function could not be revealed when there was too heavy pollution. These findings underscore that the pollutant discharge should be controlled though the natural vegetation in ecosystem promoted the purification of water bodies.

Geochemical fingerprint and spatial pattern of mine water quality in the Shaanxi-Inner Mongolia Coal Mine Base, Northwest China

The spatial distribution of mine water quality and geochemical controls must be investigated for water safety and ecosystem protection in Shaanxi-Inner Mongolian Coal Mine Base (SICMB). Based on 122 mine water samples collected from 14 mining areas, self-organizing maps (SOM) combining with principal component analysis (PCA) derived that the mine water samples were classified into seven clusters. Clusters 1 and 3 (C1 and C3) samples were dominant by HCO3-Ca and mixed types, which were distributed in the recharge area of the middle SICMB. In this area, the active groundwater circulation contributed to the good water quality. Cluster 2 (C2) samples were characterized by HCO3-Na type, mainly distributed in the discharge area of the middle SICMB. These samples were threatened by heavy fluorine contamination and high residual sodium carbonate (RSC) because of slow groundwater flow in this area. Clusters 4 and 5 (C4 and C5) samples, distributed in the northeast and middle SICMB, were characterized by high Cl- concentration and light fluorine contamination. They were influenced by anthropogenic input through faults or underground mining. In contrast, Clusters 6 and 7 (C6 and C7) samples with high salinity and sulfate were distributed in the southwest SICMB. The deep groundwater circulation enhanced water-rock interaction and contributed to poor water quality. These findings are beneficial to the management of mine water resources in the SICMB and provide an insight to investigate the mine water quality in large spatial scale.

Metabolic profile predicts incident cancer: A large-scale population study in the UK Biobank

BACKGROUND AND AIMS

Analyses to predict the risk of cancer typically focus on single biomarkers, which do not capture their complex interrelations. We hypothesized that the use of metabolic profiles may provide new insights into cancer prediction.

METHODS

We used information from 290,888 UK Biobank participants aged 37 to 73 years at baseline. Metabolic subgroups were defined based on clustering of biochemical data using an artificial neural network approach and examined for their association with incident cancers identified through linkage to cancer registry. In addition, we evaluated associations between 38 individual biomarkers and cancer risk.

RESULTS

In total, 21,973 individuals developed cancer during the follow-up (median 3.87 years, interquartile range [IQR] = 2.03-5.58). Compared to the metabolically favorable subgroup (IV), subgroup III (defined as "high BMI, C-reactive protein & cystatin C") was associated with a higher risk of obesity-related cancers (hazard ratio [HR] = 1.26, 95 % CI = 1.21 to 1.32) and hematologic-malignancies (e.g., lymphoid leukemia: HR = 1.83, 95%CI = 1.44 to 2.33). Subgroup II ("high triglycerides & liver enzymes") was strongly associated with liver cancer risk (HR = 5.70, 95%CI = 3.57 to 9.11). Analysis of individual biomarkers showed a positive association between testosterone and greater risks of hormone-sensitive cancers (HR per SD higher = 1.32, 95%CI = 1.23 to 1.44), and liver cancer (HR = 2.49, 95%CI =1.47 to 4.24). Many liver tests were individually associated with a greater risk of liver cancer with the strongest association observed for gamma-glutamyl transferase (HR = 2.40, 95%CI = 2.19 to 2.65).

CONCLUSIONS

Metabolic profile in middle-to-older age can predict cancer incidence, in particular risk of obesity-related cancer, hematologic malignancies, and liver cancer. Elevated values from liver tests are strong predictors for later risk of liver cancer.

Controlling factors and health risks of groundwater chemistry in a typical alpine watershed based on machine learning methods

Groundwater is a key water resource in alpine watersheds, but its quality is deteriorating due to human activities. The Golmud River watershed is a representative alpine watershed in Northwest China, and it was chosen to explore groundwater chemistry, associated controlling factors, source contributions, and potential health risks. The analysis includes the use of a self-organizing map (SOM), positive matrix factorization (PMF), ionic ratios, and a Monte Carlo simulation. The content of total dissolved solids in phreatic water was higher in the dry season and increased from the mountainous zone to the fine-soil plain-overflowing zone. Additionally, the water type varied from HCO3- to Cl- types whereas confined groundwater was chemically stable and of a HCO3- type. The SOM results showed a visual correlation between the ions in groundwater. The combination of SOM, PMF, and ionic ratios identified water-rock action as a dominant factor of groundwater chemistry. It was also found that Clusters I and III were mainly influenced by silicate weathering (a total contribution of 38.4 %), whereas evaporation was dominant in Cluster VI (a contribution of 32.5 %). Anthropogenic pollution was mainly associated with clusters V and IV and was related to industrial and agricultural activities during the snowmelt and wet seasons, and fluorine deposition formed by residential coal heating during the dry season (contributions of 1.4 % and 23.8 % in Clusters V and IV, respectively). The sudden increases in B3+ and Li+ in Cluster II were due to inputs from small tributaries (a contribution of 3.9 %). The probabilistic health risk assessment showed that fluoride posed a greater non-carcinogenic risk to human health than Sr2+, B3+, and NO3-, and its potential threat to children was more significant during the dry season than in other seasons. It is necessary for local governments to establish urgent fluoride emission control policies within the Golmud River watershed.

Self-organizing map improves understanding on the hydrochemical processes in aquifer systems

The holistic understanding of hydrochemical features is a crucial task for management and protection of water resources. However, it is challenging for a complex region, where multiple factors can cause hydrochemical changes in studied catchment. We collected 208 groundwater samples from such region in Kumamoto, southern Japan to explicitly characterize these processes by applying machine learning technique. The analyzed groundwater chemistry data like major cations and anions were fed to the self-organizing map (SOM) and the results were compared with classical classification approaches like Stiff diagram, standalone cluster analysis and score plots of principal component analysis (PCA). The SOM with integrated application of clustering divided the data into 11 clusters in this complex region. We confirmed that the results provide much greater details for the associated hydrochemical and contamination processes than the traditional approaches, which show quite good correspondence with the recent high resolution hydrological simulation model and aspects from geochemical modeling. However, the careful application of the SOM is necessary for obtaining accurate results. This study tested different normalization approaches for selecting the best SOM map and found that the topographic error (TE) was more important over the quantization error (QE). For instance, the lower QE obtained from min-max and log normalizations showed problems after clustering the SOM map, since the QE did not confirm the topological preservation. In contrast, the lowest TE obtained from Z-transformation data showed better spatial matching of the clusters with relevant hydrochemical characteristics. The results from this study clearly demonstrated that the SOM is a helpful approach for explicit understanding of the hydrochemical processes on reginal scale that may capably facilitate better groundwater resource management.

Fluorescence fingerprinting characteristics of water-soluble organic carbon from size-resolved particles during pollution event

A typical haze pollution process in northern China has necessitated this study which focuses on the fluorescence characteristics of water-soluble organic carbon (WSOC) in size-resolved particles. High concentrations of WSOC were found in both fine (38 μg/m³) and coarse particles (36 μg/m³) during the pollution period, which may be related to the secondary formation of organic aerosols and stable meteorological conditions. Five fluorescent components in WSOC were extracted by parallel factor analysis. Our results showed that the fluorophores in fine and coarse particles were mainly humic-like substances (humic-like, terrestrial humic-like, and high oxidation humic-like substances) and protein-like substances (protein-like and tyrosine-like substances), respectively. Moreover, the aging degree analysis, pollution source tracing, and concentration prediction of WSOC were carried out by fluorescence index. An innovative technique called self-organizing map was proposed for an in-depth investigation of the contamination mechanism of the atmospheric organic aerosol. Furthermore, the difference in the fluorescence characteristics of WSOC in fine particles was higher than that in coarse particles. The atmospheric pollution process increased the degree of difference in fluorescence characteristics. Additionally, an effective method for predicting the size of atmospheric particles was established by combining excitation-emission matrix fluorescence spectroscopy with classification and regression tree analysis.

Insight into structural composition of dissolved organic matter in saline-alkali soil by fluorescence spectroscopy coupled with self-organizing map and structural equation modeling

Soil salinization has been occurring all over the world, which severely affected crop production and threatened the life of mankind. It is necessary to take serious steps to improve soil fertility for the sustainability and productive capacity of agriculture. Soil samples of different depths were collected from native vegetation communities (Comm. Phragmites communis (CPC) and Comm. Populus alba (CPA)) and irrigated crops (corn fields (CFD) and seed melon fields (SMF)) in Hetao irrigation area of China. Three dimensional excitation-emission matrix (EEM) fluorescence technology combined with self-organizing map were used to analyze the dissolved organic matter (DOM) composition and structural characteristics in saline-alkali soils and its spatial distribution under different vegetation covers. Critical factors were recognized by classification and regression tree (CART) for distinguishing soil samples, and latent factors were revealed with structural equation modeling (SEM) for improving the humification degree of DOM from saline soils in Hetao irrigation area. Five components were obtained in the DOM substances, i.e., tyrosine-like (C1), tryptophan-like (C2), UV fulvic-like (C3), visible fulvic-like (C4) and humic-like (C5). The protein-like peaks were all obvious, and the fulvic-like peaks (600-735 a.u.) were conspicuous in the CPC soil than in others, except CFD1 and SMF1. C1 was the critical factor to distinguish native vegetation from irrigated crops, and C1 and C2 were the critical factors to distinguish CFD from SMF. Contrary to the HA/FA (0.20) and A/C (0.25), the path coefficient (-0.15) of sources with T/H was negative, indicating that the incremental contents of fluorenscense substances were in the sequences of protein-like > visible fulvic-like > UV fulvic-like > humic-like, affecting by the allochthonous. C1 (1.00) and C4 (1.00) were the primary components for improving the humification degree of DOM, which were principally originated from plant debris. EEM combined with self-organizing map, CART and SEM is an efficient way to distinguish different salinized soils and reveal the latent factors for improving the soil fertility.

Multivariate statistical analysis of potentially toxic elements in the sediments of Quanzhou Bay, China: Spatial relationships, ecological toxicity and sources identification

In this paper, the spatial distribution, pollution degree, ecological toxicity and possible sources of seven potentially toxic elements (PTEs) collected from the surface sediments of Quanzhou Bay (QZB) were analyzed by obtaining concentration measurements. The results indicated that the areas with high Cu, Pb, Zn and Hg concentrations were mainly located in the Luoyang River estuary, while the areas with high contents of Cd and As appeared in the Luoyang River estuary area and in the southern part of QZB, respectively. The contamination indices showed that the Cd pollution degree was slight to serious, while other elements were slightly enriched. The calculation results of the potential ecological risk index (RI) and toxic risk index (TRI) indicated that Cd was the main element posing ecological risk among the PTEs of sediments in QZB, followed by Hg. Moreover, in approximately 30% of the surveyed sites, PTEs exhibited low toxicity to aquatic ecosystems. Finally, the self-organizing map (SOM) and positive matrix factorization (PMF) model were used to determine the PTEs sources. Natural sources, industrial emissions, and the combustion of fossil fuels were three main sources for PTEs in the surface sediments of QZB. This study provides a reference for assessing sediment pollution and managing marine pollution in QZB.

Nexus between potentially toxic elements' accumulation and seasonal/anthropogenic influences on mangrove sediments and ecological risk in Sundarbans, Bangladesh: An approach from GIS, self-organizing map, conditional inference tree and random forest models

Mangroves play a vital role in protecting the coastal community from the climate change effect and in the restoration of the coastal ecosystem. This research has been designed to determine the spatial and seasonal changes of potentially toxic elements' (PTEs) concentration in sediments and their potential source contribution among the different human-driven processes in Sundarbans, Bangladesh. Different pollution evaluation indices, random forest (RF) model, conditional inference tree (CIT), self-organizing map (SOM), geographical information system (GIS), and principal component analysis (PCA) were used for the interpretation of sources and risk assessment of PTEs. The mean concentration of PTEs both in winter and monsoon seasons has fallen below the threshold effect level but exceeded the rare effect level of marine sediments quality standards. Results showed that the PTEs were significantly enriched (EF > 1.00 < 70.00) in sediments, whereas the Cd enrichment (7.00% samples) was very alarming (EF = 60-70). Except for Zn and Cd, other PTEs were enriched in 30-60% samples. The highest geoaccumulation and contamination factors for Cd were observed in 46-72% of samples. The ecological risk (ER) factors showed similar results where Cd showed strong to very strong factors (ER = 110-2218) in 80% of samples. The CIT explained the natural/geogenic and anthropogenic sources of pollution, where the higher CIT values for Cd indicated industrial, aquaculture, and coal-based thermal powerplant. The RF model provided that shrimp firms, power plants, industry, and seaport were recognized as the influential sources for Zn, Pb, Cr, Cd, and As in sediments. Though Pb and As were found as the most significant pollutants, Cd was identified as a severe threat to ecology and public health. Based on CIT, RF, SOM and PCA the order of PTEs in mangroves sediment were:industrial/urban > aquaculture/shrimpfirm > powerplant > seaportoperation > tourism > geogenic/natural. The present study will help the policymakers for effective and sustainable management of the mangrove ecosystem.

Light-absorption and fluorescence fingerprinting characteristics of water and methanol soluble organic compounds in PM2.5 in cold regions of Northeast China

High-performance liquid chromatography-size exclusion chromatography and excitation-emission matrix (EEM) fluorescence spectroscopy were used to analyze the seasonal variations and potential sources of molecular weight (MW) separated light-absorbing chromophores and fluorophores of water-soluble organic compounds (WSOC) and methanol-soluble organic compounds (MSOC) in PM_2.5 in cold areas of northern China. The results showed that the light-absorbing organics in MSOC had larger weight-average MW (M_w) (3.19 kDa) and number-average MW (M_n) (1.13 kDa) compared with WSOC (M_w: 1.41 kDa, M_n: 0.692 kDa). The light-absorption of organics showed a trend of winter>spring>autumn>summer and increased on air pollution days. Three fluorescent components including humic-like, protein-like, and terrestrial humic-like components in WSOC were extracted by parallel factor analysis (PARAFAC). Fluorophores in WSOC were dominated by humic-like and terrestrial humic-like components (67.7%). Three fluorescent components extracted from MSOC were low oxidation humic-like, polycyclic aromatic hydrocarbon (PAH)-like, and protein-like components respectively. It is worth noting that compared with WSOC, MSOC may have a higher human health risk due to the presence of PAH-like components. The combination of PARAFAC and self-organizing map had the potential to identify potential sources of fluorophores. It provided a new perspective for comprehensively exploring the characteristics of fluorophores in aerosols. This study provided a reference for further understanding the chemical composition and optical properties of organic aerosols in the cold regions of northern China.

Cluster-based analysis of COVID-19 cases using self-organizing map neural network and K-means methods to improve medical decision-making

In this study, we utilized unsupervised machine learning techniques to examine the relationship between different symptoms in cases who died of COVID-19 and cases who recovered from it. First, our data was cleared of redundancies, and the ten most important variables were selected using a filter-based technique (extra-tree classifier). Next, we calculated the Silhouette, Davis Boldin (DB), and the mean intra-cluster distance measures to select the optimal number of clusters, then clustered the data using both the K-means and hierarchical clustering based on Self Organizing Map (SOM) neural network. Our results revealed that patients who died of COVID-19 had high mean values in different symptoms, but not all patients with this characteristic necessarily died. Besides, our result indicated that the patient's age is directly related to the hospital duration, and elderly patients are more likely to be assigned to the intensive care unit (ICU). However, the patient's sex has the same distribution in different groups and does not correlate with other symptoms. In conclusion, our results confirmed past studies. Also, this research helps physicians improve medical services by considering other important factors for treating different groups of COVID-19 patients.

Comparative genomic analysis of the human genome and six bat genomes using unsupervised machine learning: Mb-level CpG and TFBS islands

Background

Emerging infectious disease-causing RNA viruses, such as the SARS-CoV-2 and Ebola viruses, are thought to rely on bats as natural reservoir hosts. Since these zoonotic viruses pose a great threat to humans, it is important to characterize the bat genome from multiple perspectives. Unsupervised machine learning methods for extracting novel information from big sequence data without prior knowledge or particular models are highly desirable for obtaining unexpected insights. We previously established a batch-learning self-organizing map (BLSOM) of the oligonucleotide composition that reveals novel genome characteristics from big sequence data.

Results

In this study, using the oligonucleotide BLSOM, we conducted a comparative genomic study of humans and six bat species. BLSOM is an explainable-type machine learning algorithm that reveals the diagnostic oligonucleotides contributing to sequence clustering (self-organization). When unsupervised machine learning reveals unexpected and/or characteristic features, these features can be studied in more detail via the much simpler and more direct standard distribution map method. Based on this combined strategy, we identified the Mb-level enrichment of CG dinucleotide (Mb-level CpG islands) around the termini of bat long-scaffold sequences. In addition, a class of CG-containing oligonucleotides were enriched in the centromeric and pericentromeric regions of human chromosomes. Oligonucleotides longer than tetranucleotides often represent binding motifs for a wide variety of proteins (e.g., transcription factor binding sequences (TFBSs)). By analyzing the penta- and hexanucleotide composition, we observed the evident enrichment of a wide range of hexanucleotide TFBSs in centromeric and pericentromeric heterochromatin regions on all human chromosomes.

Conclusion

Function of transcription factors (TFs) beyond their known regulation of gene expression (e.g., TF-mediated looping interactions between two different genomic regions) has received wide attention. The Mb-level TFBS and CpG islands are thought to be involved in the large-scale nuclear organization, such as centromere and telomere clustering. TFBSs, which are enriched in centromeric and pericentromeric heterochromatin regions, are thought to play an important role in the formation of nuclear 3D structures. Our machine learning-based analysis will help us to understand the differential features of nuclear 3D structures in the human and bat genomes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-022-08664-9.

Unsupervised learning methods for efficient geographic clustering and identification of disease disparities with applications to county-level colorectal cancer incidence in California

Many public health policymaking questions involve data subsets representing application-specific attributes and geographic location. We develop and evaluate standard and tailored techniques for clustering via unsupervised learning (UL) algorithms on such amalgamated (dual-domain) data sets. The aim of the associated algorithms is to identify geographically efficient clusters that also maximize the number of statistically significant differences in disease incidence and demographic variables across top clusters. Two standard UL approaches, k means with k++ initialization (k++) and the standard self-organizing map (SSOM), are considered along with a new, tailored version of the SOM (TSOM). The TSOM algorithm involves optimization of a customized objective function with terms promoting individual geographic cluster cohesion while also maximizing the number of differences across clusters, and two hyper-parameters controlling the relative weighting of geographic and attribute subspaces in a non-Euclidean distance measure within the clustering problem. The performance of these three techniques (k++, SSOM, TSOM) is compared and evaluated in the context of a data set for colorectal cancer incidence in the state of California, at the level of individual counties. Clusters are visualized via chloropleth maps and ordered graphs are also used to illustrate disparities in disease incidence among four identity groups. While all three approaches performed well, the TSOM identified the largest number of disease and demographic disparities while also yielding more geographically efficient top clusters. Techniques presented in this study are relevant to applications including the delivery of health care resources and identifying disparities among identity groups, and to questions involving coordination between county- and state-level policymakers.

Potentially toxic elemental contamination in Wainivesi River, Fiji impacted by gold-mining activities using chemometric tools and SOM analysis

Potentially toxic element (PTE) contamination in Wainivesi River, Fiji triggered by gold-mining activities is a major public health concern deserving attention. However, chemometric approaches and pattern recognition of PTEs in surface water and sediment are yet hardly studied in Pacific Island countries like Fijian urban River. In this study, twenty-four sediment and eight water sampling sites from the Wainivesi River, Fiji were explored to evaluate the spatial pattern, eco-environmental pollution, and source apportionment of PTEs. This analysis was done using an integrated approach of self-organizing map (SOM), principle component analysis (PCA), hierarchical cluster analysis (HCA), and indexical approaches. The PTE average concentration is decreasing in the order of Fe > Pb > Zn > Ni > Cr > Cu > Mn > Co > Cd for water and Fe > Zn > Pb > Mn > Cr > Ni > Cu > Co > Cd for sediment, respectively. Outcomes of eco-environmental indices including contamination and enrichment factors, and geo-accumulation index differed spatially indicated that majority of the sediment sites were highly polluted by Zn, Cd, and Ni. Cd and Ni contents can cause both ecological and human health risks. According to PCA, both mixed sources (geogenic and anthropogenic such as mine wastes discharge and farming activities) of PTEs for water and sediment were identified in the study area. The SOM analysis identified three spatial patterns, e.g., Cr-Co-Zn-Mn, Fe-Cd, and Ni-Pb-Cu in water and Zn-Cd-Cu-Mn, Cr-Ni and Fe, Co-Pb in sediment. Spatial distribution of entropy water quality index (EWQI) values depicted that northern and northwestern areas possess "poor" to "extremely poor" quality water. The entropy weights indicated Zn, Cd, and Cu as the major pollutants in deteriorating the water quality. This finding provides a baseline database with eco-environmental and health risk measures for the Wainivesi river contamination.

Potentially toxic elemental contamination in Wainivesi River, Fiji impacted by gold-mining activities using chemometric tools and SOM analysis

Potentially toxic element (PTE) contamination in Wainivesi River, Fiji triggered by gold-mining activities is a major public health concern deserving attention. However, chemometric approaches and pattern recognition of PTEs in surface water and sediment are yet hardly studied in Pacific Island countries like Fijian urban River. In this study, twenty-four sediment and eight water sampling sites from the Wainivesi River, Fiji were explored to evaluate the spatial pattern, eco-environmental pollution, and source apportionment of PTEs. This analysis was done using an integrated approach of self-organizing map (SOM), principle component analysis (PCA), hierarchical cluster analysis (HCA), and indexical approaches. The PTE average concentration is decreasing in the order of Fe > Pb > Zn > Ni > Cr > Cu > Mn > Co > Cd for water and Fe > Zn > Pb > Mn > Cr > Ni > Cu > Co > Cd for sediment, respectively. Outcomes of eco-environmental indices including contamination and enrichment factors, and geo-accumulation index differed spatially indicated that majority of the sediment sites were highly polluted by Zn, Cd, and Ni. Cd and Ni contents can cause both ecological and human health risks. According to PCA, both mixed sources (geogenic and anthropogenic such as mine wastes discharge and farming activities) of PTEs for water and sediment were identified in the study area. The SOM analysis identified three spatial patterns, e.g., Cr-Co-Zn-Mn, Fe-Cd, and Ni-Pb-Cu in water and Zn-Cd-Cu-Mn, Cr-Ni and Fe, Co-Pb in sediment. Spatial distribution of entropy water quality index (EWQI) values depicted that northern and northwestern areas possess "poor" to "extremely poor" quality water. The entropy weights indicated Zn, Cd, and Cu as the major pollutants in deteriorating the water quality. This finding provides a baseline database with eco-environmental and health risk measures for the Wainivesi river contamination.

How can users' comments posted on social media videos be a source of effective tags?

This paper proposed a new approach for the extraction of tags from users' comments made about videos. In fact, videos on the social media, like Facebook and YouTube, are usually accompanied by comments where users may give opinions about things evoked in the video. The main challenge is how to extract relevant tags from them. To the best of the authors' knowledge, this is the first research work to present an approach to extract tags from comments posted about videos on the social media. We do not pretend that comments can be a perfect solution for tagging videos since we rather tried to investigate the reliability of comments to tag videos and we studied how they can serve as a source of tags. The proposed approach is based on filtering the comments to retain only the words that could be possible tags. We relied on the self-organizing map clustering considering that tags of a given video are semantically and contextually close. We tested our approach on the Google YouTube 8M dataset, and the achieved results show that we can rely on comments to extract tags. They could be also used to enrich and refine the existing uploaders' tags as a second area of application. This can mitigate the bias effect of the uploader's tags which are generally subjective.

Machine learning in modelling land-use and land cover-change (LULCC): Current status, challenges and prospects

Land-use and land-cover change (LULCC) are of importance in natural resource management, environmental modelling and assessment, and agricultural production management. However, LULCC detection and modelling is a complex, data-driven process in the remote sensing field due to the processing of massive historical and current data, real-time interaction of scenario data, and spatial environmental data. In this paper, we review principles and methods of LULCC modelling, using machine learning and beyond, such as traditional cellular automata (CA). Then, we examine the characteristics, capabilities, limitations, and perspectives of machine learning. Machine learning has not yet been dramatic in modelling LULCC, such as urbanization prediction and crop yield prediction because competition and transition between land cover types are dynamic at a local scale under varying natural drivers and human activities. Upcoming challenges of machine learning in modelling LULCC remain in the detection and prediction of LULC evolutionary processes if considering their applicability and feasibility, such as the spatio-temporal transition mechanisms to describe occurrence, transition, spreading, and spatial patterns of changes, availability of training data of all the change drivers, particularly sequence data, and identification and inclusion of local ecological, hydrological, and social-economic drivers in addressing the spectral feature change. This review points out the need for multidisciplinary research beyond image processing and pattern recognition of machine learning in accelerating and advancing studies of LULCC modelling. Despite this, we believe that machine learning has strong potentials to incorporate new exploratory variables in modelling LULCC through expanding remote sensing big data and advancing transient algorithms.

Source apportionment and ecological and health risk mapping of soil heavy metals based on PMF, SOM, and GIS methods in Hulan River Watershed, Northeastern China

Heavy metals in agricultural soils not only affect the food security and soil security, but also endanger the human health through the food chain. Based on the incorporation of index analysis, positive matrix factorization (PMF), self-organizing map (SOM), and geostatistical methods, this research performed the assessment of source apportionment and ecological and health risks of soil heavy metals in Hulan River Watershed, Northeastern China. According to the Pollution Load Index (PLI), 83.08% of the soil samples were slightly or mildly polluted, and 1.54% of the soil samples were severely polluted. The ecological risk index (EI) showed that about 80.77% and 60.77% of the soil samples were beyond the low risk level for Hg and Cd, respectively. In this research, the non-carcinogenic and carcinogenic risk indices for children were higher than adult males and adult females. Four potential sources were revealed based on the PMF and SOM analysis including atmospheric deposition and industrial emission; transportation source; agricultural source; and a combination of agricultural, industrial, and natural sources. Considerable and high ecological risk from Hg existed in the area close to the coal steam-electric plant, and considerable and high ecological risk from Cd existed in the Hulan River estuary area. The eastern part of the study area experienced higher non-carcinogenic and carcinogenic risks for adults and children than the western part of the study area. The source apportionment and ecological and health risk mapping provide important role in reducing pollution sources. Zonal pollution control and soil restoration measures should be performed in the areas with high ecological and health risks.

Collaborative Mining of Whole Genome Sequences for Intelligent HIV-1 Sub-Strain(s) Discovery

BACKGROUND

Effective global antiretroviral vaccines and therapeutic strategies depend on the diversity, evolution, and epidemiology of their various strains as well as their transmission and pathogenesis. Most viral disease-causing particles are clustered into a taxonomy of subtypes to suggest pointers toward nucleotide-specific vaccines or therapeutic applications of clinical significance sufficient for sequence-specific diagnosis and homologous viral studies. These are very useful to formulate predictors to induce cross-resistance to some retroviral control drugs being used across study areas.

OBJECTIVE

This research proposed a collaborative framework of hybridized (Machine Learning and Natural Language Processing) techniques to discover hidden genome patterns and feature predictors, for HIV-1 genome sequences mining.

METHOD

630 human HIV-1 genome sequences above 8500 bps were excavated from the National Center for Biotechnology Information (NCBI) database (https://www.ncbi.nlm.nih.gov) for 21 countries across different continents, Antarctica exempt. These sequences were transformed and learned using a self-organizing map (SOM). To discriminate emerging/new sub-strain(s), the HIV-1 reference genome was included as part of the input isolates/samples during the training. After training the SOM, component planes defining pattern clusters of the input datasets were generated, for cognitive knowledge mining and subsequent labelling of the datasets. Additional genome features including dinucleotide transmission recurrences, codon recurrences, and mutation recurrences, were finally extracted from the raw genomes to construct output classification targets for supervised learning.

RESULTS

SOM training explains the inherent pattern diversity of HIV-1 genomes as well as inter- and intra-country transmissions in which mobility might play an active role, as corroborated by literature. Nine sub-strains were discovered after disassembling the SOM correlation hunting matrix space attributed to disparate clusters. Cognitive knowledge mining separated similar pattern clusters bounded by a certain degree of correlation range, discovered by the SOM. A Kruskal-Wallis rank-sum test and Wilcoxon rank-sum test showed statistically significant variations in dinucleotide, codon, and mutation patterns.

CONCLUSION

Results of the discovered sub-strains and response clusters visualizations corroborate existing literature, with significant haplotype variations. The proposed framework would assist in the development of decision support systems for easy contact tracing, infectious disease surveillance, and studying the progressive evolution of the reference HIV-1 genome.

Taxon-specific sensitivities to flow intermittence reveal macroinvertebrates as potential bioindicators of intermittent rivers and streams

As complex mosaics of lotic, lentic, and terrestrial habitats, intermittent rivers and ephemeral streams (IRES) support high biodiversity. Despite their ecological importance, IRES are poorly represented in routine monitoring programs, but recent recognition of their considerable-and increasing-spatiotemporal extent is motivating efforts to better represent IRES in ecological status assessments. We examine response patterns of aquatic macroinvertebrate communities and taxa to flow intermittence (FI) across three European climatic regions. We used self-organizing map (SOM) to ordinate and classify sampling sites based on community structure in regions with continental, Mediterranean and oceanic climates. The SOM passively introduced FI, quantified as the mean annual % flow, and visualized its variability across classified communities, revealing a clear association between community structure and FI in all regions. Indicator species analysis identified taxa indicative of low, intermediate and high FI. In the continental region, the amphipod Niphargus was indicative of high FI and was associated with groundwater-fed IRES, whereas indicators of Mediterranean IRES comprised Odonata, Coleoptera and Heteroptera taxa, which favor lentic conditions. In the oceanic region, taxa indicative of relatively high FI included leuctrid stoneflies and a limnephilid caddisfly, likely reflecting the colonization of IRES by aerial adults from nearby perennial reaches. The Diptera families Chironomidae and Simuliidae showed contrasting FI preferences among regions, reflecting environmental heterogeneity between regions and the coarse taxonomic resolution to which these organisms were identified. These region-specific community and taxon responses of aquatic biota to FI highlight the need to adapt standard biotic indices to enable effective ecological status assessments in IRES.

Ligand Based Virtual Screening Using Self-organizing Maps

Conventional drug discovery methods rely primarily on in-vitro experiments with a target molecule and an extensive set of small molecules to choose the suitable ligand. The exploration space for the selected ligand being huge; this approach is highly time-consuming and requires high capital for facilitation. Virtual screening, a computational technique used to reduce this search space and identify lead molecules, can speed up the drug discovery process. This paper proposes a ligand-based virtual screening method using an artificial neural network called self-organizing map (SOM). The proposed work uses two SOMs to predict the active and inactive molecules separately. This SOM based technique can uniquely label a small molecule as active, inactive, and undefined as well. This can reduce the number of false positives in the screening process and improve recall; compared to support vector machine and random forest based models. Additionally, by exploiting the parallelism present in the learning and classification phases of a SOM, a graphics processing unit (GPU) based model yields much better execution time. The proposed GPU-based SOM tool can successfully evaluate a large number of molecules in training and screening phases. The source code of the implementation and related files are available at https://github.com/jayarajpbalakrishnan/2_SOM_SCREEN.

Characterizing land use effect on shallow groundwater contamination by using self-organizing map and buffer zone

Nitrate-nitrogen (NO₃-N) contamination in groundwater is a major problem of drinking and domestic waters in rural areas. This study revealed the influence of land use type on shallow alluvial groundwaters in a typical rural area in South Korea by applying a self-organizing map (SOM), principal component analysis (PCA), and hierarchical cluster analysis (HCA). The uncertainty of spatial information on land use was improved by using a buffer zone of the average influence radius of 32.65 m surrounding wells. Two major land-use types, forests (44.9%) and rice fields (28.8%), occupied a total of 73.7% of the rural area. The higher concentrations of NO₃-N in public facilities and livestock areas were demonstrated to directly recharge groundwater pollutants. NO₃-N contamination in rice paddies, which also contained chlorine (Cl) and sulfate (SO₄), was assessed according to the nutrients and residual salt in the soil. In addition, different NO₃-N concentrations for the same land use indicate various biochemical reactions and NO₃-N recharge types into the groundwater system. The shallow groundwaters in the study area were classified into three clusters according to their chemical constituents and land-use properties, especially NO₃-N concentration, including pH, Cl, and SO₄, using a SOM, PCA, and HCA. Unlike existing studies, we applied a buffer zone based on the Cooper-Jacob equation to obtain an improved SOM model prediction accuracy approximately 10% greater than that using the original dataset.

A dynamic analysis of household debt using a self-organizing map

The Korean consumer credit panel offers a well-organized set of microdata representing various characteristics of individual borrowers. To overcome the difficulty of fragmented microdata details, we construct a cluster of Korean consumers' credit, to develop a self-organizing map that visualizes individuals' characteristics along two dimensions. The result of cluster analysis reveals that most borrowers belong to one large cluster representing diligent borrowers who honor their loan payments. Conversely, several small clusters that represent borrowers with high default probability are identified, and we also found that these borrowers' characteristics vary. No significant change is found in the structure of the cluster, even when the aggregate amount of consumer credit is increased. Moreover, the expansionary monetary policy did not change the quantitative structure of household debt in Korea.

SUPPLEMENTARY INFORMATION

The online version contains supplementary material available at 10.1007/s00181-021-02120-5.

Approximate spectral clustering using both reference vectors and topology of the network generated by growing neural gas

Spectral clustering (SC) is one of the most popular clustering methods and often outperforms traditional clustering methods. SC uses the eigenvectors of a Laplacian matrix calculated from a similarity matrix of a dataset. SC has serious drawbacks: the significant increases in the time complexity derived from the computation of eigenvectors and the memory space complexity to store the similarity matrix. To address the issues, I develop a new approximate spectral clustering using the network generated by growing neural gas (GNG), called ASC with GNG in this study. ASC with GNG uses not only reference vectors for vector quantization but also the topology of the network for extraction of the topological relationship between data points in a dataset. ASC with GNG calculates the similarity matrix from both the reference vectors and the topology of the network generated by GNG. Using the network generated from a dataset by GNG, ASC with GNG achieves to reduce the computational and space complexities and improve clustering quality. In this study, I demonstrate that ASC with GNG effectively reduces the computational time. Moreover, this study shows that ASC with GNG provides equal to or better clustering performance than SC.

Comparison of multimodal findings on epileptogenic side in temporal lobe epilepsy using self-organizing maps

OBJECTIVE

To develop a decision-making tool to evaluate and compare the performance of neuroimaging markers with clinical findings and the significance of attributes for presurgical lateralization of mesial temporal lobe epilepsy (mTLE).

METHODS

Thirty-five unilateral mTLE patients who qualified as candidates for surgical resection were studied. Seizure semiology, ictal EEG, ictal epileptogenic zone, interictal-irritative zone, and MRI findings were used as clinical markers. Hippocampal T1 volumetry and FLAIR intensity, DTI estimated; mean diffusivity (MD) in the hippocampus and fractional anisotropy (FA) in posteroinferior cingulum and crus of fornix, and the output of logistic regression method on volumetrics of the hippocampus, amygdala, and thalamus were adopted as neuroimaging markers. The self-organizing map (SOM) method was applied to markers to provide predictive methods for mTLE lateralization.

RESULTS

The SOM clustered all clinical attributes correctly with 100% accuracy and sensitivity for both the left and right mTLE. Among the clinical markers, seizure semiology and interictal-irrelative zone are the most sensitive attribute for the left-mTLE group lateralization. The accuracy achieved by applying the SOM method to the neuroimaging attributes was 94%, while the sensitivity was achieved 90% for left and 100% for right mTLE. SOM evidence indicated that the hippocampal volume is the most sensitive attribute for the prediction of the laterality in left-mTLE groups.

CONCLUSION

The proposed SOM method showed that neuroimaging markers may not replace with clinical findings. Nevertheless, multimodal neuroimaging can play an effective role in preoperative lateralization to reduce the costs and risks of surgical resection.

Oral health, organic and inorganic saliva composition of men with Schizophrenia: Case-control study

BACKGROUND

Schizophrenia (SCZ) presents complex challenges related to diagnosis and clinical monitoring. The study of conditions associated with SCZ can be facilitated by using potential markers and patterns that provide information to support the diagnosis and oral health.

METHODS

The salivary composition of patients diagnosed with SCZ (n = 50) was evaluated and compared to the control (n = 50). Saliva samples from male patients were collected and clinical parameters were evaluated. The concentration of total proteins and amylase were determined and salivary macro- and microelements were quantified by ICP OES and ICP-MS. Exploratory data analysis based on artificial intelligence tools was used in the investigation.

RESULTS

There was a significant increase in the salivary concentrations of Al, Fe, Li, Mg, Na, and V, higher prevalence of caries (p < 0.001), periodontal disease (p < 0.001), and reduced salivary flow rate (p = 0.019) in SCZ patients. Also, samples were grouped into six clusters. As, Co, Cr, Cu, Mn, Mo, Ni, Se, and Sr were correlated with each other, while Fe, K, Li, Ti, and V showed the highest concentrations in the samples distributed in the clusters with the highest association between SZC patients and controls.

CONCLUSIONS

The results obtained indicate changes in salivary flow, organic composition, and levels of macro- and microelements in SCZ patients. Salivary concentrations of Fe, Mg, and Na may be related to oral conditions, higher prevalence of caries, and periodontal disease. The exploratory analysis showed different patterns in the salivary composition of SCZ patients impacted by associations between oral health conditions and the use of medications. Future studies are encouraged to confirm the results investigated in this study.

Self-organizing map applied to the choice of internal standards for the determination of Cd, Pb, Sn, and platinum group elements by inductively coupled plasma mass spectrometry

The behaviors of internal standards, according to different flow rates of the cell collision gas (He), were studied for the determination of Cd, Pb, Pd, Pt, Rh, and Sn in samples of fish and mollusks by inductively coupled plasma mass spectrometry (ICP-MS). The elements Bi, Ge, In, Sc, and Y were selected as internal standards, considering their masses and first ionization energies. Addition and recovery experiments were carried out at three concentration levels to evaluate the accuracy of the method applied for the analysis of two samples with different matrices. The results were evaluated using a self-organizing map (SOM). The best analyte/IS pairs were as follows: ¹¹⁴Cd⁺/⁷⁴Ge⁺, ¹⁹⁵Pt⁺/⁷⁴Ge⁺, and ²⁰⁸Pb⁺/⁷⁴Ge⁺. For ¹⁰³Rh⁺, ¹⁰⁶Pd⁺, and ¹²⁰Sn⁺, greater accuracy was achieved without use of an internal standard. Helium gas (2.8 mL min^-1) was used in the collision cell for the analytes, except for Sn, and recoveries ranged from 98 to 101% under optimal conditions. The use of SOM as an exploratory analysis tool was an effective approach for selection of the most appropriate internal standards.

Assessment of extrinsic and intrinsic influences on water quality variation in subtropical agricultural multipond systems

Understanding wetland water quality dynamics and associated influencing factors is important to assess the numerous ecosystem services they provide. We present a combined self-organizing map (SOM) and linear mixed-effects model (LMEM) to relate water quality variation of multipond systems (MPSs, a common type of non-floodplain wetlands in agricultural regions of southern China) to their extrinsic and intrinsic influences for the first time. Across the 6 test MPSs with environmental gradients, ammonium nitrogen (NH₄⁺-N), total nitrogen (TN), and total phosphate (TP) almost always exceeded the surface water quality standard (2.0, 2.0, and 0.4 mg/L, respectively) in the up- and midstream ponds, while chlorophyll-a (Chl-a) exhibited hypertrophic state (≥28 μg/L) in the midstream ponds during the wet season. Synergistic influences explained 69±12% and 73±10% of the water quality variations in the wet and dry season, respectively. The adverse, extrinsic influences were generally 1.4, 6.9, 3.2, and 4.3 times of the beneficial, intrinsic influences for NH₄⁺-N, nitrate nitrogen (NO₃^--N), TP, and potassium permanganate index (COD_Mn), respectively, although the influencing direction and degree of forest and water area proportion were spatiotemporally unstable. While COD_Mn was primarily linked with rural residential areas in the midstream, higher TN and TP concentrations in the up- and midstream were associated with agricultural land, and NH₄⁺-N reflected a small but non-negligible source of free-range poultry feeding. Pond surface sediments exhibited consistent, adverse effects with amplifications during rainfall, while macrophyte biomass can reflect the biological uptake of COD_Mn and Chl-a, especially in the mid- and downstream during the wet season. Our study advances nonpoint source pollution (NPSP) research for small water bodies, explores nutrient "source-sink" dynamics, and provides a timely guide for rural planning and pond management. The modelling procedures and analytical results can inform refined assessment of similar NFWs elsewhere, where restoration efforts are required.

Emerging POPs-type cocktail signatures in Pusa caspica in quantitative structure-activity relationship of Caspian Sea

The Caspian seal Pusa caspica is the only endemic mammalian species throughout the Caspian Sea. This is the first report on risk assessment of persistent organic pollutants (POPs) in Caspian seals by age-sex and tissue-specific uptake, and their surrounding environment (seawater, surface sediments, and suspended particulate matters, SPMs) in the Gorgan Bay (Caspian Sea, Iran). Among the quantified 70 POPs (∑₃₅PCBs, ∑₃HCHs, ∑₆CHLs, ∑₆DDTs, ∑₁₇PCDD/Fs, HCB, dieldrin, and aldrin), ∑₃₅PCBs were dominant in abiotic matrices (48.80% of ∑₇₀POPs), followed by HCHs > CHLs > DDTs > PCDD/Fs > other POPs in surface sediments > SPMs > seawater, while the toxic equivalent quantity (TEQ_WHO) exceeded the safe value (possible risk in this area). In biota, the highest levels of ∑₇₀POPs were found in males (756.3 ng g^-1 dw, p < 0.05), followed by females (419.0 ng g^-1 dw) and pups (¹91.6 ng g^-1 dw) in liver > kidney > muscle > blubber > intestine > fur > heart > spleen > brain. The positive age-related POPs declining correlation between mother-pup pairs suggested the possible maternal transfer of POPs to offspring. The cocktail toxicity assessment revealed that Caspian seals can pose a low risk based on their mixed-TEQ values. Self-organizing map (SOM) indicated the non-coplanar PCB-93 as the most over-represented functional congener in tissue-specific POPs bioaccumulation. Quantitative toxicant tissue-profiling is valuable for predicting the state of mixture toxicity in pinniped species.

Characterizing the anthropogenic-induced trace elements in an urban aquatic environment: A source apportionment and risk assessment with uncertainty consideration

The spatial distribution of water quality status, especially in water bodies near intensively urbanized areas, is tightly associated with patterns of human activities. For establishing a robust assessment of the sediment quality in an urban aquatic environment, the source apportionment and risk assessment of Cr, Mn, Ni, Cu, Zn, As, Cd, Hg, and Pb in sediments from an anthropogenic-influenced lake were carried out with considering uncertainties from the analysis methods, random errors in the sample population and the spatial sediment heterogeneity. The distribution analysis of the trace metals with inverse distance weighting-determined method showed that the pollutants were concentrated in the middle and southern areas of the lake. According to the self-organizing map and constrained positive matrix factorization receptor model, agricultural sources (24.8%), industrial and vehicular sources (42.5%), and geogenic natural sources (32.7%) were the primary contributors to the given metals. The geogenic natural had the largest random errors, but the overall result was reliable according to the uncertainty analysis. Furthermore, the stochastic contamination and ecological risk models identified a moderate/considerable contamination level and a moderate ecological risk to the urban aquatic ecosystem. With consideration of uncertainties from the spatial heterogeneity, the contamination level of Hg, and the ecological risk of Cd in had a 20-30% probability of the increase.

Pixel precise unsupervised detection of viral particle proliferation in cellular imaging data

Cellular and molecular imaging techniques and models have been developed to characterize single stages of viral proliferation after focal infection of cells in vitro. The fast and automatic classification of cell imaging data may prove helpful prior to any further comparison of representative experimental data to mathematical models of viral propagation in host cells. Here, we use computer generated images drawn from a reproduction of an imaging model from a previously published study of experimentally obtained cell imaging data representing progressive viral particle proliferation in host cell monolayers. Inspired by experimental time-based imaging data, here in this study viral particle increase in time is simulated by a one-by-one increase, across images, in black or gray single pixels representing dead or partially infected cells, and hypothetical remission by a one-by-one increase in white pixels coding for living cells in the original image model. The image simulations are submitted to unsupervised learning by a Self-Organizing Map (SOM) and the Quantization Error in the SOM output (SOM-QE) is used for automatic classification of the image simulations as a function of the represented extent of viral particle proliferation or cell recovery. Unsupervised classification by SOM-QE of 160 model images, each with more than three million pixels, is shown to provide a statistically reliable, pixel precise, and fast classification model that outperforms human computer-assisted image classification by RGB image mean computation. The automatic classification procedure proposed here provides a powerful approach to understand finely tuned mechanisms in the infection and proliferation of virus in cell lines in vitro or other cells.

•

Automatic classification of cell imaging data for in vitro viral propagation in host cells.

•

Economical in terms of computation times for rapid change/no change detection in large image data prior to human decision making.

•

Simulating viral proliferation/cell recovery by progressive and selective single pixel changes in contrast polarity and/or intensity.

Assessment of water quality classes using self-organizing map and fuzzy C-means clustering methods in Ergene River, Turkey

Surface water is one of the primary sources for drinking, irrigation, and industrial activities in Ergene River, Turkey. However, its quality has deteriorated due to the point and non-point pollution sources. Therefore, an appropriate assessment of surface water quality is very important. Water quality classification is calculated separately for each quality parameter in Turkey. An overall assessment of surface water quality is essential for water management. In this study, self-organizing maps (SOMs) and fuzzy C-means clustering (FCM) methods have been used for assessing surface water quality in the Ergene River. Seven water quality parameters have been considered as important indicators to evaluate water quality status in 7 observation points located in the river, covering the period from 1985 to 2013.

Building a stable and accurate model for heavy metal detection in mulberry leaves based on a proposed analysis framework and laser-induced breakdown spectroscopy

Laser-induced breakdown spectroscopy (LIBS) was used to rapidly detect heavy metals in mulberry leaves. For the purpose of increasing detection stability and accuracy, a novel analysis framework consisting of a Kohonen self-organizing map (SOM), a variable selection method using the successive projection algorithm (SPA) and uninformative variable elimination (UVE), and a consensus modeling strategy was proposed for processing LIBS data to determine copper (Cu) and chromium (Cr) content. Results showed that the best regression model for Cu and Cr content achieved the residual predictive deviation (RPD) values of 10.0494 and 8.3874, respectively, and root mean square error of prediction (RMSEP) values of 110.4550 and 41.4561, respectively. The proposed strategy provides a high-accuracy and rapid alternative to the traditional method for monitoring heavy metals in mulberry leaves, which could guarantee the quality of mulberry leaves and potentially be used in food-related industries.