Innovating Drug Design for Alzheimer’s Disease via Reinforcement Learning for Enhanced Molecular Generation

Alzheimer’s disease is a neurodegenerative disorder that affects the brain’s intricate functions. With an estimated 50 million people worldwide affected by Alzheimer’s disease, innovative approaches to drug discovery are imperative. Developing drugs for Alzheimer's disease is challenging due to the intricacies of the brain and prolonged clinical trial durations. These challenges underscore the critical need for innovative approaches in the pursuit of effective Alzheimer's treatments. In this study, a Random Forest predictor is utilized to refine features for predictive modeling. Employing a stacked Gated Recurrent Unit architecture in the generative phase, diverse Simplified Molecular Input Line Entry System (SMILES) format molecular drugs are produced. Reinforcement learning optimizes the model for higher bioactivity. This approach yields robust outcomes, with an 84% validity rate for generated SMILES and a Tanimoto similarity coefficient of 0.70 to known molecules. This methodology showcases promising results, emphasizing its efficacy in computational drug discovery for Alzheimer’s disease.