Heavy metal(loid)s (HMs) play an important role in economic development since they are used in various branches of industry. However, all industrial activities emit HMs into the environment, where they are no longer useful but potentially toxic. It has been observed that HMs enter the soil and sediment, and potential human health risk may arise due to their excessive accumulation. Having in mind the importance of assessing the risk due to HMs in these media, we analyzed published works in the last decade and created a summary of exposure factors in health risk models for HMs in soil and sediment. This analysis revealed a remarkable increase in the number of publications about health risk assessment of HMs, especially in the last few years. Since many differences in the values of the exposure factors and their distributions were noted, this study focused on elucidating these differences. Non-carcinogenic and carcinogenic health risk assessment models were evaluated through a deterministic approach that is prevalent in use, and a probabilistic one, which is gaining more and more attention in research. In the end, guided by the studied literature, we propose the values and distributions for the exposure factors.

This study investigated the spatial distribution of HMs in agricultural and urban soils near the largest oil refining complex in Serbia, identified pollution sources, and assessed ecological and human health risks. A large fraction of soil samples showed elevated Hg (40% of samples), Pb (53%), Cd (90%), and As (93%) concentrations compared to the background levels. Hotspots for Pb, As, Hg, Cd, and Zn were observed in the industrial area, indicating significant anthropogenic input. Multivariate analysis, including PMF, revealed four contamination sources: emissions from the oil refining industry, agricultural activities, traffic emissions, and natural background. The pollution indices mostly fell into the moderate pollution range, with As, Hg, and Cd showing the highest enrichment. The potential ecological risk index (RI) indicated that about one-third of the samples had moderate ecological risk and determined a major RI hotspot near the refinery. The health risk assessment identified As and Cr as the largest contributors to non-carcinogenic risk, although the average HI was below one. Monte Carlo simulation confirmed that adults and children had negligible health risks at the 95th percentile and highlighted exposure frequency and body weight as the most influential exposure parameters. Based on source-specific risk, the oil refining industry emissions had the highest impact on HI and TCR values.

A rapid and efficient ultrasound-assisted extraction (UAE) procedure followed by inductively coupled plasma mass spectrometry (ICP-MS) was developed for the determination of 14 rare earth elements (REEs) (La, Ce, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu), along with yttrium (Y) and scandium (Sc), in coffee samples. The method was validated using certified reference material (NIST SRM 1547), recovery tests at four fortification levels, and comparisons with microwave-assisted digestion (MAD). Excellent accuracy and precision were achieved, with recovery rates ranging from 80.1% to 112% and relative standard deviations (RSD%) below 14%. Limits of detection (LODs) ranged from 0.2 ng/kg (Yb) to 0.16 µg/kg (Nd). Total REE concentrations varied between 8.3 µg/kg and 1.1 mg/kg, with the highest individual mean concentrations (µg/kg) observed for Ce (11.7), La (6.0), and Sc (4.7). The lowest individual mean concentrations (µg/kg) were for Ho (0.16), Lu (0.066), and Tm (0.063). Multivariate analysis of REE profiles from 92 coffee samples collected in Serbia revealed clear distinctions between ground roasted and instant coffees, as well as between different surrogate blends. This study indicated that the determination of coffee’s geographical origin was not possible due to the diverse types, blends, and additives. However, differences in REE profiles suggest potential classification based on variety. REEs pose a negligible health risk to coffee consumers, with HI values ranging from 4.7 × 10⁻⁸ to 6.3 × 10⁻⁶ and TCR ranging from 2.6 × 10⁻¹⁴ to 3.5 × 10⁻¹².

Accurate prediction of the Langelier Saturation Index (LSI), an indicator of water’s scaling and corrosive potential, is vital for water treatment and infrastructure maintenance. In this study, five machine learning models (Ridge Regression, Support Vector Machine, Random Forest, Deep Neural Network, and XGBoost) were applied to predict the LSI from physicochemical characteristics of groundwater in the Morava River basin (Serbia). Rigorous data preprocessing (outlier removal, missing data handling, z-score normalization) and feature selection were performed to ensure robust model training. Models were optimized via 10-fold cross-validation on a 70/30 train–test split. All models achieved high predictive accuracy, with ensemble methods outperforming others. XGBoost yielded the best performance (R² = 0.98; RMSE = 0.06), followed closely by Random Forest (R² = 0.95). The linear Ridge model showed the lowest (yet still strong) performance (R² = 0.90) and larger errors at extreme LSI values. Feature importance analysis consistently identified pH as the most influential predictor of the LSI, followed by alkalinity and calcium. Partial dependence plots confirmed that the models captured established nonlinear LSI behavior. The LSI rises steeply with increasing pH and moderately with mineral content. Overall, this comparative study demonstrates that modern machine learning models can predict the LSI accurately, providing interpretable insights through feature importance and dependence plots. These results underscore the potential of data-driven approaches to complement traditional water stability indices for proactive water quality management.

The degradation of sediments in urban environments worldwide is driven by population growth, urbanization, and industrialization, highlighting the need for thorough quality assessment and management strategies. As a result of these anthropogenic activities, benzene, toluene, ethylbenzene, xylenes, and styrene (BTEXS) are persistently released into the environment, polluting sediment. This study employed self-organizing maps (SOMs), positive matrix factorization (PMF), and Monte Carlo simulation of source-oriented health risks to comprehensively investigate sediment in an urban shallow lake in a mid-sized city in central Serbia. The results indicated a mean ∑BTEXS concentration of 225 µg/kg, with toluene as the dominant congener, followed by m,p-xylene, benzene, ethylbenzene, o-xylene, and styrene. Three contamination sources were identified: waste solvents and plastic waste due to intensive recreational activities, and vehicle exhaust from heavy traffic surrounding the lake. Both non-carcinogenic and carcinogenic health risks were below the permissible limits. However, children were more susceptible to health risks. Benzene from vehicle exhaust is the most responsible for non-carcinogenic and carcinogenic health risks in both population groups. The results of this study can help researchers to find a suitable perspective on the dynamics and impacts of BTEXS in lake sediments.