基于机器学习的亚砷酸印迹聚合物功能单体-靶标结合性能预测模型OA

The widespread arsenic contamination in water bodies has become a global concern,driving the need for high-performance detection materials.Traditional experimental methods are often time-consuming and resource-intensive,while molecular simulations are limited by high computational costs and low throughput when handling large candidate libraries.To address this,we developed machine learning models to predict the binding energy(ΔE)between arsenite and molecularly imprinted polymer functional monomers.Using 13 molecular descriptors-including structural and quantum chemical features such as hydrogen bond donor count,heavy atom count,and molecular orbital energies-we trained random forest,least squares boosting,and eXtreme gradient boosting models,optimized via bayesian optimization and sparrow search algorithm.The XGBoost model tuned with sparrow search algorithm performed best,achieving R² values of 0.98672(training)and 0.94938(test),with root mean square error(RMSE)and mean absolute error(MAE)as low as 0.14304 and 0.06506,respectively.Shapley additive exPlanations(SHAP)analysis identified total molecular energy and dipole moment as the most influential features.External validation yielded an R² of 0.71820,confirming the model's generalization capability.This work provides a data-driven strategy for the rational design of functional monomers in arsenite-imprinted polymers.