Scaling, performance, and quality: Machine-Learning evidence for how WWTP and compost processes shape microplastics

Composting is a key pathway for microplastics (MPs) to enter soils via organic waste. Based on 635 records from 99 studies, MP abundance was higher in sludge-based and mixed composts than in plant-derived materials. Meta-analysis of wastewater treatment plants (WWTPs) found that primary clarification removed the most MPs. Among secondary treatments, MBR sludge had the highest MP levels - 68.53% higher than SBR and 5.6% higher than CAS. The machine-learning models (random forest with SHAP and partial dependence) were used to resolve actionable process levers with R² > 0.72 for WWTP process technologies and >0.77 for scenario prediction. Two parameters dominated: total plastic removal rate (TPR) and daily treatment capacity (DTC), with an operational window (TPR ≈ 92.5-96% and DTC <300,000 m³/d) minimizing downstream MP burdens. Finally, Scenario projections suggest two mitigation pathways: source control to reduce influent loads and performance-based upgrades to sludge handling.