Eco-friendly insulating gases are essential for the green upgrading of power transmission and distribution systems. Although predictive modeling of insulation performance and physicochemical properties has been widely studied

biosafety

as a critical practical indicator

is governed by distinct structural factors and requires independent predictive models. In this work

following OECD guidelines

we developed the first QSAR models for the acute inhalation toxicity of fluorocarbon-based eco-friendly insulating gases. Through performance evaluation

support vector machine with radial basis function kernel (SVM-RBF) and eXtreme gradient boosting(XGBoost) were selected as two top-performing models

which demonstrated excellent performance in training set fitting capability

cross-validation stability

and external prediction generalizability. The SHapley additive explanations(SHAP) analysis was further applied to interpret the relationships between key molecular descriptors and toxicity

revealing the core structural features influencing biosafety. The results provide a theoretical basis and analytical framework for the molecular design and risk control of fluorocarbon-based eco-friendly insulating gases.