Natural Products in AI-Driven Drug Discovery: A Bibliometric and Thematic Review of Computational Methods, Disease Areas, and Emerging Frontiers

Download PDF

Natural products remain a major source of chemically diverse scaffolds for drug discovery, while artificial intelligence has created new opportunities to mine, represent, prioritize, and interpret complex natural product chemical space. This original bibliometric and thematic review examines how AI-driven natural product drug discovery is organized as a research field, with attention to computational methods, disease-area priorities, data infrastructures, emerging frontiers, and translational barriers. The review applies bibliometric logic by separating search strategy, dataset construction, eligibility criteria, metadata cleaning, keyword normalization, research-cluster interpretation, and thematic synthesis. Because verified database-export counts were not available for direct computation in the present execution, numerical bibliometric indicators, ranking metrics, keyword frequencies, and cluster-size claims are not reported. Instead, the analysis maps the approved evidence base through structured charting and cautious thematic interpretation. The review focuses on natural products, phytochemicals, medicinal plants, microbial and marine natural products where supported, natural product databases, machine learning, deep learning, molecular representation learning, network pharmacology, virtual screening, target prediction, ADMET and toxicity prediction, and natural product-inspired molecular design. Disease-area mapping is interpreted cautiously across cancer, infectious disease, inflammatory, metabolic, neurodegenerative, antimicrobial, and antiviral discovery themes where the mapped evidence supports discussion. The synthesis identifies a field shaped by database quality, molecular-standardization needs, representation learning, target and mechanism inference, screening prioritization, explainability, generative design, and validation constraints. The review contributes a research agenda for more reproducible, interpretable, experimentally validated, and translationally responsible AI-enabled natural product discovery.