Application of machine learning and multivariate approaches for assessing microplastic pollution and its associated risks in the urban outdoor environment of Bangladesh.

Microplastics (MPs) are an emerging global concern due to severe toxicological risks for ecosystems and public health. Therefore, this is the first study in Bangladesh to assess MP pollution and its associated risks for ecosystems and human health in the outdoor urban environment using machine learning and multivariate approaches. The occurrences of MPs in the urban road dust were 52.76 ± 20.24 particles/g with high diversity, where fiber shape (77%), 0.1-0.5 mm size MPs (75%), blue color (26%), and low-density polyethylene (24%) polymer was the dominating MPs category. Pollution load index value (1.28-4.42), showed severe pollution by MPs. Additionally, the contamination factor (1.00-5.02), and Nemerow pollution index (1.38-5.02), indicate moderate to severe MP pollution. The identified polymers based on calculated potential ecological risk (2248.52 ± 1792.79) and polymer hazard index (814.04 ± 346.15) showed very high and high risks, respectively. The occurrences of MPs could effectively be predicted by random forest, and support random vector machine, where EC, salinity, pH, OC, and texture classes were the influencing parameters. Considering the human health aspect, children and adults could be acutely exposed to 19259.68 and 5777.90 MP particles/ year via oral ingestion. Monte-Carlo-based polymers associated cancer risk assessment results indicate moderate risk and high risk for adults and children, respectively, where children were more vulnerable than adults for MP pollution risks. Overall assessment mentioned that Dhaka was the most polluted division among the other divisions.

Removal of hazardous textile dye from simulated wastewater by municipal organic solid waste charcoal using machine learning approaches: Kinetics, isotherm, and thermodynamics.

This study focuses on the probable use of municipal organic solid waste charcoal (MOSWC) as an adsorbent for Methyl orange (MO) adsorption. The prepared MOSWC is characterized by FE-SEM and FT-IR. Batch adsorption experiments were conducted with the influencing of different operational conditions namely time of contact (1-180 min), adsorbate concentration (60-140 mg/L), adsorbent dose (1-5 g/L), pH (3-11), and temperature (25-60 °C). The high coefficient value (R2 = 0.96) of the process optimization model suggests that this model was significant, where pH and adsorbent dose expressively stimulus adsorption efficiency including 40.11 mg/g at pH (3), MO concentration (100 mg/L), and MOSWC dose (1 g/L). Furthermore, the machine learning approaches (ANN and BB-RSM) revealed a good association between the tested and projected values. The highest monolayer adsorption capacity of MO was 90.909 mg/g. Pseudo-second-order was the well-suited kinetics, where Langmuir isotherm could explain better for equilibrium adsorption data. Thermodynamic study shows MO adsorption is favourable, exothermic, and spontaneous. Finally, this study indicates that MOSWC could be a potential candidate for the adsorption of MO from wastewater.

Black carbon derived PET plastic bottle waste and rice straw for sorption of Acid Red 27 dye: Machine learning approaches, kinetics, isotherm and thermodynamic studies.

This study focuses on the probable use of PET waste black carbon (PETWBC) and rice straw black carbon (RSBC) as an adsorbent for Acid Red 27 (AR 27) adsorption. The prepared adsorbent is characterized by FE-SEM and FT-IR. Batch adsorption experiments were conducted with the influencing of different operational conditions namely time of contact (1-180 min), AR 27 concentration (5-70 mg/L), adsorbent dose (0.5-20 g/L), pH (2-10), and temperature (25-60°C). High coefficient value [PETWBC (R2 = 0.94), and RSBC (R2 = 0.97)] of process optimization model suggesting that this model was significant, where pH and adsorbent dose expressively stimulus removal efficiency including 99.88, and 99.89% for PETWBC, and RSBC at pH (2). Furthermore, the machine learning approaches (ANN and BB-RSM) revealed a good association between the tested and projected value. Pseudo-second-order was the well-suited kinetics, where Freundlich isotherm could explain better equilibrium adsorption data. Thermodynamic study shows AR 27 adsorption is favourable, endothermic, and spontaneous. Environmental friendliness properties are confirmed by desorption studies and satisfactory results also attain from real wastewater experiments. Finally, this study indicates that PETWBC and RSBC could be potential candidates for the adsorption of AR 27 from wastewater.