Encoding Chemical Intelligence: A Sustainable Multi-Objective Drug Prediction and Optimization Using Quantum_D-CART Method

In rare diseases, drug discovery remains computationally intensive because of the various search spaces in molecules and limited clinical information. This research proposes a Quantum Drug Context-Aware Reinforced Transformer (Quantum_D-CART) method, which is a innovation of context-aware reinforcement with unified transformer (CASAR and UTF) in improvised drug discovery. The study uses the SMILES molecular dataset for encoding the structures of molecules, physicochemical properties such as num_atoms,$\operatorname{logP}$and pIC50, and represents the unified interaction. The enhanced reinforced loop over quantum optimizes the selection of binding pathways and exploration. The outcome demonstrates that the Quantum_D-CART method achieved the greatest prediction accuracy pIC50 of 92.5%, outperforming the traditional methods, including 83.4% of DeepChem, 84.9% MolGAN, and 86.7 % of ChemBERTa. The suggested method also achieved faster convergence and reduced exploration rate of 43 % and 48 % respectively. The obtained results show the robustness of the proposed model under several circumstances, like data scarcity, exploration rate, and better prediction in binding affinity. For drug discovery in complex diseases, this method is scalable in providing a quantum-accelerated solution and adaptive treatment strategies.