Towards eco-friendly and biodegradable pesticides: intelligent consensus modelling and read-across for predicting soil half-life

Pesticides are widely used in agriculture to enhance crop yield and protect against pests. However, their persistence in soil can lead to long-term environmental contamination and pose health risks to humans and other organisms through indirect exposure via the food chain. In this study, we used in silico approaches like Quantitative Structure-Activity Relationship (QSAR) modelling, Intelligent Consensus Prediction (ICP), and chemical read-across to predict the soil degradation half-lives of various pesticides. Models were established using 2D molecular descriptors, thoroughly validated with the help of training and test sets validation parameters, and conformed to OECD guidelines. The predictive models were applied to the Pesticide Properties Database (PPDB) to demonstrate their utility in screening untested and/or newly synthesized pesticides, considering the domain of applicability. Key structural features associated with degradation were identified, providing valuable insights for the design of biodegradable and environmentally safer pesticides. This work contributes to data gap-filling, regulatory risk assessment, and the prioritization of new or untested pesticides for environmental evaluation.