Presentation (CTI)

OBJECTIVES: Drug repurposing offers a cost-effective strategy to identify novel therapeutic applications for existing drugs. However, potential repurposing signals are usually dispersed within vast heterogeneous literature, often appearing as secondary endpoints, subgroup analyses, or exploratory outcomes challenging to detect during manual review. This study aimed to develop an AI-assisted, structured literature review for rapid detection of such signals, using iron supplementation as a proof-of-concept to demonstrate feasibility.
METHODS: An AI-assisted systematic literature review was conducted using a predefined protocol incorporating eligibility criteria, search strategy, and comprehensive documentation, with human reviewer validation at every stage. PubMed search retrieved 2,140 records. Following AI-driven deduplication and screening with human oversight, 46 studies (anemia = 13; non-anemia = 33) were included for qualitative synthesis. AI-assisted extraction was applied to identify off-label therapeutic benefits with findings reviewed and interpreted by subject-matter experts.
RESULTS: Across multiple clinical indications, iron supplementation exhibited several off-label therapeutic benefits including improvements in fertility, fatigue, quality-of-life, exercise capacity, postoperative recovery, cardiopulmonary function, and psychological outcomes. Reductions in hospital stay, infection risk, transfusion needs, and disease-related hospitalizations were also noted. In cardiovascular and chronic kidney disease populations, iron therapy was associated with improved ventricular function, WHO functional status, and quality-of-life, with signals suggesting reduced hospitalization and mortality in specific settings. Ferric carboxymaltose emerged as a prominent formulation showing consistent functional and healthcare utilization benefits across diverse populations. Reported risks were generally limited and aligned with established safety profiles. The review process was completed in 84 hours using AI-assistance with an AI screening accuracy of 85%.
CONCLUSIONS: This proof-of-concept study demonstrates the feasibility and rapid utility of AI-assisted literature review methods for identifying early drug-repurposing signals from existing evidence. AI-driven detection can complement conventional review methodologies, facilitating systematic, reproducible hypothesis generation to inform downstream clinical and translational research, while maintaining human oversight and interpretability.