Hierarchical discrimination of pyrethroid types and varieties using a ceria nanozyme array with machine learning

Area of Science:

• Environmental Chemistry
• Analytical Chemistry
• Materials Science

Background:

• Pyrethroid pesticides (PYRs) pose environmental risks, with toxicity linked to their chemical structure, specifically the presence or absence of a cyano group (Type I/II).
• Existing rapid detection methods lack the capability for hierarchical identification of PYRs, from broad class discrimination to specific compound identification.

Purpose of the Study:

• To develop a method for hierarchical identification of pyrethroid pesticides, enabling discrimination from class level (Type I/II) down to individual compounds.
• To create a sensitive and accurate detection system for PYRs in various environmental and food matrices.

Main Methods:

• Screening of cerium-based nanozymes with aromatic carboxylic acid ligands to identify those responsive to PYRs.
• Construction of a four-channel colorimetric array utilizing differential nanozyme responses to PYRs.
• Application of a multi-output random forest (Mul-RF) model for hierarchical classification and concentration prediction.

Main Results:

• The nanozyme array successfully achieved hierarchical identification of PYRs, distinguishing between Type I/II and eight specific varieties, independent of concentration.
• The Mul-RF model accurately predicted PYR concentrations after initial classification.
• Validation in spiked spinach, soil, lake water, and cotton hulls demonstrated no cross-category misclassification or variety confusion.

Conclusions:

• The developed nanozyme array combined with machine learning offers a new paradigm for precise risk assessment of pyrethroid pesticides.
• This strategy enables on-site, stepwise screening of PYRs in environmental and food samples, improving detection capabilities.
• The hierarchical identification approach enhances the accuracy and reliability of pesticide analysis.