Decoding the Blood-Brain Barrier Permeation in CNS Drug Discovery: An Integrated Framework through High Fidelity Explainable ML and Low Dimensional Linear Discriminant Mapping

our -anchored set (n = 154), our mechanismaligned LDA yields a direct, interpretable discriminant score (DF > 0.5) without surrogate explanations and maintains-aligned LDA yields a direct, interpretable discriminant score (DF > 0.5) without surrogate explanations and maintains performance on an external 741compound cohort, supporting broader -compound cohort, supporting broader chemicalspace-space generalization Our framework integrates prediction, interpretation, and prioritization to guide CNS medicinal chemistry workflow and provides a clear link between modelling and design of molecules.