AI-Assisted Drug Discovery from Phytoconstituents: Predicting Drug–Target Interactions

Artificial intelligence (AI) has emerged as a powerful computational tool to support early-stage drug discovery by facilitating the analysis and prioritization of bioactive compounds. In the context of plant-derived phytoconstituents, AI-based methods offer significant potential to assist drug–target interaction (DTI) prediction, virtual screening, and pharmacological prioritization. This narrative review summarizes recent advances in AI-assisted approaches applied to phytochemical-based drug discovery, with a particular focus on machine learning, deep learning, and graph-based models used for DTI prediction. The review also discusses the role of AI in supporting in-silico assessment of absorption, distribution, metabolism, excretion, and toxicity (ADME/Tox) properties, as well as its integration with molecular docking- and pharmacology-oriented evaluation workflows. A structured literature search was conducted using PubMed, Scopus, Web of Science, and Google Scholar, covering publications from 2015 to 2025, with the final literature search completed on 15 December 2025 and employing keywords related to artificial intelligence, drug–target interactions, phytochemicals, and natural product drug discovery. Rather than reporting new experimental findings, this review critically analyzes existing computational strategies, databases, and workflows, highlighting their strengths, limitations, and translational relevance. Overall, AI-assisted DTI prediction is presented as a complementary approach that can guide experimental pharmacology and accelerate the rational development of phytoconstituent-based therapeutics, provided that computational predictions are supported by subsequent in-vitro and in-vivo validation. This review is narrative in scope and is based exclusively on previously published studies, without generating new experimental or computational data.