A Predictive ML‐QSAR Approach for Targeting GSK‐3β in Alzheimer's Disease

Abstract Alzheimer’s disease (AD) is a progressive neurological disorder marked by cognitive decline. Glycogen Synthase Kinase‐3 beta (GSK‐3β), a key enzyme, promotes AD by causing tau protein hyper phosphorylation. Effective GSK‐3β inhibitors are urgently needed. This research employed a Machine Learning‐Based Quantitative Structure‐Activity Relationship (ML‐QSAR) strategy to discover new inhibitors. Using 510 known inhibitors, predictive models—Artificial Neural Network (ANN), Support Vector Machine (SVM), and Random Forest (RF)—were developed. A major result was the superior performance of the PubChem fingerprint‐based ANN model, which achieved an excellent correlation coefficient of 0.9743 on the training set and 0.8059 on the test set, demonstrating high predictive accuracy. This optimized model then screened central nervous system (CNS) and targeted libraries, using Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET) filtering. This process identified 33 promising hit candidates with predicted high potency (pIC50 ≥ 7). A key finding from molecular docking and dynamics simulations confirmed F3393‐0263 as the top candidate, showing stable binding and high affinity with GSK‐3β. This compound emerges as a novel, potential therapeutic agent for AD treatment.