Chemical Repeated-Dose Toxicity Data with OECD Test Guideline Validation

A data-driven modeling study for predicting subchronic oral repeated-dose toxicity of chemicals. The dataset, compiled from eChemPortal and J-CHECK databases, contains study-derived effect levels (NOAEL and LOAEL) for model development. Models were validated experimentally and applied to screen the Pesticides Properties DataBase.

Use Cases

• Predicting chemical toxicity based on NOAEL and LOAEL effect levels mentioned in the description
• Screening pesticide databases for potential repeated-dose toxicity hazards as described in the study
• Identifying toxic structural motifs, such as chlorine- and amine-group-containing aromatic systems, via substructure analysis
• Developing regulatory screening tools based on models trained on OECD guideline-compliant data

Strengths

• All underlying studies followed standardized OECD test guidelines (407, 408, 422) under GLP conditions, ensuring regulatory relevance.
• Final models were experimentally validated using substances from Global Product Compliance (GPC) Europe AB.
• Model test set accuracies ranged from 0.642 to 0.682, with training set accuracies up to 0.902.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is small in scale (429.6 KB), indicating a limited scope of chemical coverage.

Provenance

Source

eChemPortal and J-CHECK databases.

Collection Method

Data compiled for machine learning model development following OECD guidelines.

Freshness

Last updated 2026-03-25 18:03:46; freshness should be verified.