What question did this study set out to answer?

Systematic evaluation of data preprocessing and model selection strategies for reliable pIC 50 prediction of acetylcholinesterase inhibitors

View Full Paper

Ask AI

Bookmark

Authors

Evaluation of machine learning strategies predicts AChE activity, indicating challenges in dataset correlation.

The aim is to evaluate different data preprocessing and model selection strategies for predicting AChE inhibitory activity.
Utilized machine learning models including CatBoost, XGBoost, and Support Vector Regression.
Applied data preprocessing strategies such as logarithmic transformation and winsorization of IC50 values.
Implemented a 70-15-15 train-validation-test split and 10-fold cross validation for performance assessment.
Explored stacking based ensemble learning strategies to enhance model generalization.
Optimized tree-based models achieved the highest accuracy for predicting AChE activity.
Predictive performance was influenced more by intrinsic dataset characteristics than by model selection.
Stacking models provided marginal improvements over the best individual learners.
SHAP analysis revealed important contributions from molecular descriptors, aiding interpretability.

Delibaş et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark

View Full Paper

Ask AI

Bookmark

Authors

Evaluation of machine learning strategies predicts AChE activity, indicating challenges in dataset correlation.

The aim is to evaluate different data preprocessing and model selection strategies for predicting AChE inhibitory activity.
Utilized machine learning models including CatBoost, XGBoost, and Support Vector Regression.
Applied data preprocessing strategies such as logarithmic transformation and winsorization of IC50 values.
Implemented a 70-15-15 train-validation-test split and 10-fold cross validation for performance assessment.
Explored stacking based ensemble learning strategies to enhance model generalization.
Optimized tree-based models achieved the highest accuracy for predicting AChE activity.
Predictive performance was influenced more by intrinsic dataset characteristics than by model selection.
Stacking models provided marginal improvements over the best individual learners.
SHAP analysis revealed important contributions from molecular descriptors, aiding interpretability.

Delibaş et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark