Artificial intelligence driven innovation in tuberculosis drug discovery

Artificial intelligence (AI) is rapidly transforming tuberculosis (TB) drug discovery by addressing major limitations of traditional pipelines, such as prolonged timelines, excessive cost, and challenges imposed by the biology of Mycobacterium tuberculosis (Mtb). Through machine learning (ML) and deep learning (DL), researchers can systematically analyze vast, heterogeneous biological and chemical datasets, leading to more efficient identification of novel drug targets, accelerated virtual screening, and predictive modeling of anti-TB activity. AI-driven methodologies now enhance multiple stages of drug development: from target and hit identification, to de novo drug design through generative modeling, and into preclinical optimization and clinical trial design. These technologies facilitate the identification of candidates active against both drug-sensitive and drug-resistant Mtb, enable prioritization of compounds with desirable pharmacokinetic and safety profiles, and advance the integration of resistance genomics for durable therapeutic strategies. Nonetheless, persistent challenges including limited model generalizability, lack of mechanistic insight, data quality issues, and the need for rigorous, prospective validation continue to hinder transformative progress. This review surveys the current landscape of AI-powered TB drug research, critically comparing it with traditional approaches, highlighting both advances and limitations, and outlining future opportunities at the intersection of computational innovation and TB therapeutics.