Enhancing Blood-Brain Barrier Penetration Prediction by Machine Learning-Based Integration of Novel and Existing, In...

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Pharmacology

Background:

• Predicting blood-brain barrier (BBB) penetration is critical for central nervous system (CNS) drug development, yet poses a significant challenge.
• Existing prediction models often lack standardization and geometric optimization, leading to calculation variations, particularly for polar surface area (PSA).

Purpose of the Study:

• To develop a novel machine learning (ML)-based scoring system for BBB penetration prediction.
• To create a standardized database and novel in silico 3D calculation for PSA.
• To evaluate and compare the performance of the ML model against existing prediction rules.

Main Methods:

• Developed a standardized dataset of 154 molecules with 24 molecular parameters, including experimental and in silico measurements.
• Implemented a novel in silico 3D calculation for nonclassical PSA.
• Trained and validated ML models using a 100-fold Monte Carlo cross-validation framework.
• Employed explainable artificial intelligence (AI) methods like Shapley additive explanations (SHAP) for parameter influence analysis.

Main Results:

• The ML random forest model achieved superior performance in predicting binary BBB penetration (AUC 0.88) and multiclass predictions (AUC 0.82).
• The ML approach significantly outperformed existing scores like CNS MPO (AUC 0.53) and BBB score (AUC 0.68).
• SHAP analysis confirmed the multifactorial nature of BBB penetration and the advantage of multivariate ML models.

Conclusions:

• The integrated ML approach enhances BBB penetration prediction accuracy by combining experimental and in silico data with novel methods.
• This approach can reduce reliance on extensive experimental measurements and animal testing, accelerating CNS drug development.
• The study highlights the potential of ML and explainable AI in optimizing drug discovery for CNS-related therapies.