Quantum Gradient Optimized Drug Repurposing Prototype for Omics Data
Document Type
Article
Publication Title
arXiv
Publication Date
6-23-2025
Abstract/ Summary
This paper presents a novel quantum-enhanced prototype for drug repurposing and addresses the challenge of managing massive genomics data in precision medicine. Leveraging cutting-edge quantum server architectures, we integrated quantum-inspired feature extraction with large language model (LLM)–-based analytics and unified high-dimensional omics datasets and textual corpora for faster and more accurate therapeutic insights. Applying Synthetic Minority Over-sampling Technique (SMOTE) to balance underrepresented cancer subtypes and multi-omics sources such as TCGA and LINCS, the pipeline generated refined embeddings through quantum principal component analysis (QPCA). These embeddings drove an LLM trained on biomedical texts and clinical notes, generating drug recommendations with improved predicted efficacy and safety profiles. Combining quantum computing with LLM outperformed classical PCA-based approaches in accuracy, F1 score, and area under the ROC curve. Our prototype highlights the potential of harnessing quantum computing and next-generation servers for scalable, explainable, and timely drug repurposing in modern healthcare.