Quantum Machine Learning-Driven Drug Repurposing Framework for Emerging Infectious Diseases Using Biomedical Knowledge Graphs and Multi-Omics Data

Abstract

Background: To ensure adequate health of adolescents, emerging infectious diseases are a real menace, but the traditional method of drug discovery is a 10-15-year process that demands huge financial resources. Drug repurposing using artificial intelligence is an alternative that may be promising, yet there is a lack of biological knowledge integration and the use of multimodal data. Purpose: The proposed research is based on the Quantum Machine Learning (QML)-based drug repurposing framework that allows combining the biomedical knowledge graph with the multi-omics data to determine the effective therapeutic candidates to new infectious diseases in adolescents. Methods: The data were a simulated biomedical graph consisting of 6,000 multi-omics profiles, 5,200 approved drugs, 18,000 genes, 9,500 proteins, and 42,000 drug-target interactions. A hybrid quantum Variational learning framework was used to couple graph embedding with multi-omics representations to rank drugs, whereas therapeutic recommendations were interpreted using a SHAP-based explainability. Results: The proposed framework demonstrated 96.4% accuracy, 95.8% precision, 96.1% recall, 95.9% F1-score, 0.985 ROC-AUC, 0.981 PR-AUC, which is 4.7-8.9 higher than the conventional graph neural network and deep learning baselines in terms of evaluation metrics. The model also achieved 92.8 percent Top-10 drug recommendation accuracy, and an average of 0.23 s per patient inference time. Conclusion: The suggested QML concept illustrates the future prospects of using quantum learning, biomedical knowledge graphs, and multi-omics data to expedite explainable drug repurposing and assist precise therapeutic decisions in emerging infectious issues in adolescent groups.