Harnessing waste-derived amendments for microbially mediated acid mine drainage remediation: from mechanisms to process optimization and machine learning

Acid mine drainage (AMD) remains a persistent environmental challenge, creating a need for treatment strategies that combine sustained contaminant control with resource-efficient use of residual materials. Waste-derived amendments may couple residue valorization with microbially mediated treatment of AMD, particularly mine-tailings leachate, but direct evidence combining variable waste-derived donors with heterogeneous collected leachate remains limited, and machine-learning (ML) models have rarely been validated across amendment batches, reactors, or sites. This review therefore evaluates amendment functions, microbial and geochemical mechanisms, operational constraints, and the role of ML in process development. A targeted narrative synthesis was conducted using publisher webpages, PubMed, and institutional repositories through 17 September 2026. Evidence was organized into four application settings: collected tailings-leachate treatment, source-zone tailings amendment, other real mine waters, and synthetic AMD; modeling studies were assessed separately for transferability. Representative studies show that performance is strongly endpoint-dependent. In 460-day field reactors containing organic substrates, sulfate removal was 26–35 mol% and Zn removal reached 88–95 mol%, whereas Mn removal was only 2–6 mol%. In a 172-day ethanol-fed reactor treating successive tailings leachates, sulfate removal reached 84.7 ± 6.9%, most metal removals approached 100%, and Mn removal was approximately 80%. A 17-run brewery-wastewater modeling study reported within-study R 2 values of 0.97 and 0.99, demonstrating predictive feasibility but not transferable or autonomous control. The review is limited by selective rather than exhaustive searching, incomplete access to some full texts, and heterogeneity in influent chemistry, treatment boundaries, and sampling periods, which precluded pooled analysis. Overall, waste-derived amendments can support sulfate reduction, alkalinity generation, biomass retention, and metal partitioning, but strong metal removal can coexist with incomplete sulfate conversion and element-specific limitations. Future progress requires continuous and field-scale validation, matched mass balances, disturbance testing, residual-stability assessment, and ML used as constrained decision support rather than as a substitute for mechanistic and prospective operating evidence.