CAESAR models for developmental toxicity

Area of Science:

• Toxicology
• Computational Chemistry
• Regulatory Science

Background:

• The European REACH legislation mandates extensive chemical safety assessments.
• Developmental toxicity testing is complex, time-consuming, and expensive.
• Quantitative Structure-Activity Relationship (QSAR) methods are encouraged by REACH for chemical assessment.

Purpose of the Study:

• To develop QSAR models for predicting developmental toxicity.
• To create user-friendly tools for chemical risk assessment under REACH.
• To minimize false negatives in developmental toxicity predictions.

Main Methods:

• Development of two distinct QSAR models for developmental toxicity.
• Utilized random forest and adaptive fuzzy partition algorithms.
• Implementation of one model into the CAESAR online application.

Main Results:

• Two QSAR models for developmental toxicity were successfully developed.
• Both models demonstrated strong performance.
• The random forest model was integrated into the CAESAR online application for public access.

Conclusions:

• CAESAR QSAR models are designed to minimize false negatives, enhancing their utility for REACH compliance.
• The CAESAR online application provides accessible tools for industry and regulators.
• The developed models facilitate easier assessment of developmental toxicity and other endpoints.