Integrated Machine-Learning Framework for Balanced Performance-Safety Design of Iron-Based

A literature-derived dataset of 1007 cases supports an integrated machine-learning framework for environmental remediation. The data encompasses 80 iron-based materials, 136 target pollutants, and 50 test organisms. Author Qi Chen developed this framework to quantitatively co-assess the performance and safety of remediation materials.

Use Cases

• Training binary QSAR models for remediation performance prediction based on molecular descriptors.
• Predicting the direction of material-pollutant joint toxicity based on organismal attributes.
• Ranking candidate iron-based remediation materials using a dominance-aware dynamic weighting scheme.
• Applying feature attribution analysis to identify key drivers like material heterogeneity for performance.

Strengths

• Dataset contains 1007 cases derived from literature, providing a substantial foundation for model training.
• Model predictions achieved balanced accuracies of 0.88 for remediation performance and 0.90 for joint toxicity.
• Framework includes an applicability-domain filter to limit overconfident extrapolation.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect publication bias inherent to literature-derived sources.

Provenance

Source

figshare

Collection Method

Derived from literature review.

Freshness

Last updated 2026-06-04 20:30:02; freshness should be verified.