Integrating Artificial Intelligence Into Drug Discovery From Medicinal Plants: Current Applications and Infrastructural Challenges

Medicinal plants represent a vast and evolutionarily refined reservoir of structurally diverse bioactive compounds that have historically contributed to major therapeutic breakthroughs. However, despite their pharmacological richness, systematic translation of plant-derived metabolites into clinically approved drugs remains constrained by persistent bottlenecks, including extract complexity, dereplication redundancy, structural elucidation challenges, taxonomic ambiguity, and multi-component pharmacology. This review presents a bottleneck-driven and systems-oriented framework for integrating artificial intelligence (AI) into medicinal plant-based drug discovery (MPDD) to address some of these bottlenecks. Rather than reiterating broadly documented AI tools used in synthetic drug development, the manuscript critically examines phytomedicine-specific challenges and maps AI applications across key stages of the pipeline: medicinal plant identification, extraction optimization, plant metabolite identification and dereplication, absorption, distribution, metabolism, excretion, and toxicity (ADMET) prediction, virtual screening, network pharmacology, repurposing, and generative de novo design of pseudo-natural products. Emphasis is placed on the foundational requirement of making phytochemical datasets truly AI-ready through metadata harmonization, dataset balancing, novelty-aware modeling, and structured data engineering. Emerging approaches such as transformer-based foundational models, graph neural networks (GNNs), generative AI (GAI) architectures, and multi-omics-integrated network pharmacology are discussed within a pragmatic translational context. Importantly, this review maintains a balanced perspective, acknowledging that fully AI-driven clinically approved botanical drugs/synthetic drugs have yet to emerge and identifies challenges limiting the progress. By integrating computational methods with infrastructural reform, this work outlines a progressive roadmap to transition AI in medicinal plant research from exploratory studies toward standardized, reproducible, and biologically grounded next-generation drug discovery.