Multivariate modeling of engineered nanomaterial features associated with developmental toxicity.

Area of Science:

• Environmental Science
• Toxicology
• Materials Science

Background:

• Engineered nanomaterials (ENMs) are increasingly used in consumer products.
• Understanding ENM toxicity is crucial due to their widespread application.
• The influence of physicochemical characteristics (PCC) on ENM behavior and toxicity is not fully understood.

Purpose of the Study:

• To investigate the relationship between ENM PCC and developmental toxicity.
• To develop a predictive model for ENM toxicity using novel PCC parameters.
• To utilize an embryonic zebrafish model for comprehensive toxicity assessment.

Main Methods:

• Analysis of ENM PCC not previously studied in developmental toxicity.
• Testing a diverse panel of ENMs using an embryonic zebrafish model.
• Application of machine learning (ML) to characterize combinatorial PCC sets and predict toxicity.

Main Results:

• No singular PCC could accurately predict ENM bioactivity, indicating nonlinear relationships.
• Combinatorial sets of PCC, including concentration, surface area, shape, and polydispersity, were found to be key predictors.
• The ML model successfully captured the developmental toxicity profile of ENMs, considering whole-organism effects.

Conclusions:

• Physicochemical characteristics are critical determinants of ENM developmental toxicity.
• A machine learning approach integrating multiple PCC provides a robust framework for predicting ENM toxicity.
• This study advances the understanding of ENM safety assessment for consumer products.