Predicting Acute Oral Toxicity in Bobwhite Quail: Development of QSAR Models for LD50
2025
Nadia Iovine | Alessandra Roncaglioni | Emilio Benfenati
The development of a predictive model for estimating oral acute toxicity (LD50) in wildlife species is essential for environmental risk assessments. In this study, a quantitative structure&ndash:activity relationship (QSAR) model was developed to predict the acute oral toxicity of pesticides toward Bobwhite quail, categorizing them into three toxicity classes: low, moderate, and high. This model was built using the SARpy softwareA dataset of pesticides collected from OpenFoodTox and the ECOTOX database was used to identify training and test datasets, while data collected from the PPDB were used as an external validation. The model&rsquo:s performance was evaluated using these three sets. The accuracy achieved on the training set was 0.75, indicating good performance during model development. However, the model&rsquo:s accuracy dropped to 0.55 for the test set, suggesting some overfitting. The external validation accuracy was 0.69, reflecting the model&rsquo:s ability to generalize to new, unseen data. While these results demonstrate the potential of the QSAR models for predicting toxicity in Bobwhite quail, they also highlight the need for further refinement to improve predictive accuracy, particularly for unseen compounds. This work contributes to the development of computational tools for wildlife risk assessment and toxicological predictions.
显示更多 [+] 显示较少 [-]