Reprogramming Cancer Therapeutics through Artificial Intelligence-Guided Drug Repurposing: Computational Innovation, Experimental Validation, and Translational Challenges

Drug repurposing, which involves identifying the anticancer activity of drugs initially developed for non-oncologic diseases, has shown particular promise in oncology. Artificial intelligence (AI), with its capacity to rapidly combine chemical, biological, clinical, and real-world evidence and generate novel hypotheses and predictions beyond traditional discovery paradigms, has the potential to transform the drug repurposing process in oncology. Here, we provide an overview of the translational challenges in AI-based predictions for cancer drug repurposing. A PubMed and Google Scholar literature search was conducted using Medical Subject Headings and Boolean logic operators related to drug repurposing and artificial intelligence or machine learning. Peer-reviewed, English-language literature describing AI-enabled approaches to drug repurposing was primarily considered, with an emphasis on foundational work and, when possible, on the last five years. Particular focus was placed on literature that provided preclinical or clinical validation. The reviewed literature indicates that AI can facilitate a variety of overlapping repurposing strategies, including new drug–target pair identification, candidate prioritization, drug combination synergy prediction, and patient-level opportunity stratification. Areas of increasing focus include toxicity-aware repurposing modeling, enhanced interpretability, multi-omics molecular signature integration, and formulation/delivery optimization guided by AI. In summary, AI-based repurposing is an evolving, cost-effective, and promising approach for advancing precision oncology, and it will require greater emphasis on data quality, biological validation, and model interpretability to enable the translation of computational predictions into real-world impact.