Machine learning predicts drug passage through blood brain barrier accurately
An Explainable Machine Learning Framework for Predicting Blood-Brain Barrier Permeability Using Molecular Descriptors
Machine Learning
Summary
Drugs need to cross the blood-brain barrier (BBB) to treat brain diseases, but predicting which drugs can do this is hard. The authors developed a computer method that uses chemical features of molecules to guess if they can pass the BBB. They tested four algorithms and found that one called XGBoost worked best, predicting correctly about 89% of the time. They also explained which molecular features mattered most, helping to understand how the predictions were made.
What this means in practice
- •For pharmaceutical researchers: Screen early-stage drug candidates for likely ability to cross the blood-brain barrier using an interpretable machine learning model.
- •For chemical data scientists: Use the explainable framework to identify key molecular features influencing blood-brain barrier permeability for compound design.
Authors
Fatemeh Mahmoudi
Abstract
Blood-brain barrier (BBB) permeability is a critical determinant in the development of central nervous system therapeutics because it directly influences the ability of drug candidates to reach their target sites within the brain. In this study, an explainable machine learning framework was developed to predict BBB permeability using molecular descriptors generated from the MoleculeNet BBBP dataset with the RDKit cheminformatics toolkit. Fifteen physicochemical descriptors extracted from 2,039 compounds were used to train four supervised machine learning algorithms, including Logistic Regression, Support Vector Machine (SVM), Random Forest, and Extreme Gradient Boosting (XGBoost). Hyperparameter optimization was performed using GridSearchCV, while model interpretability was investigated using SHapley Additive exPlanations (SHAP). Among the evaluated models, the optimized XGBoost classifier achieved the best predictive performance, with an accuracy of 88.97%, a precision of 88.92%, a recall of 97.76%, an F1-score of 93.13%, and a ROC-AUC of 0.9282. Stratified five-fold cross-validation further demonstrated the robustness of the proposed model, yielding a mean ROC-AUC of 0.8982 +/- 0.0130. Feature importance and SHAP analyses consistently identified TPSA, HBD, and LogP as the most influential molecular descriptors governing BBB permeability prediction. Overall, the proposed framework provides an accurate, interpretable, and computationally efficient approach for BBB permeability prediction and may serve as a valuable tool for the early-stage screening of CNS drug candidates.