Abstract C005: An AI platform for systematic identification of drug repurposing candidates for rare pediatric cancers via the FDA 505(b)(2) regulatory pathway

Abstract

Rare pediatric cancers collectively represent a substantial burden of childhood cancer mortality, yet their individually small patient populations deter conventional drug development investment. Many drugs with established activity against pediatric tumor biology are approved internationally but unavailable in the United States, or are off-patent generics lacking a commercial sponsor for FDA-approved pediatric oncology indications. The FDA 505(b)(2) regulatory pathway, which permits reliance on published literature and prior findings of safety and efficacy, offers an accelerated route to approval for such candidates when paired with orphan drug and rare pediatric disease designations that provide market exclusivity and Priority Review Voucher eligibility. We developed a multi-stage AI platform that systematically screens off-patent drugs against rare pediatric cancers and generates the regulatory documentation required to advance candidates toward FDA submission. The platform evaluates drug-disease combinations across regulatory status, clinical evidence from indexed literature and trial registries, GMP-compliant supply chain availability, and designation eligibility. Candidates meeting initial thresholds undergo AI-powered deep-dive analysis using large language models with full-text literature retrieval, producing structured clinical evidence packages. A commercial validation layer generates market intelligence, competitive landscape analysis, and intellectual property strategy assessments. The platform incorporates a regulatory documentation engine that produces FDA Orphan Drug Designation applications and development strategy packages with built-in anti-hallucination safeguards and citation verification. The platform has been validated through successful FDA interactions, including cleared Investigational New Drug applications and granted designations for rare pediatric conditions. A computational chemistry module extends the platform by generating novel prodrug modifications, enabling composition-of-matter patent protection for otherwise unpatentable off-patent molecules. This technology transforms the economic calculus for rare pediatric cancer drug development by reducing the time and cost of identifying, validating, and advancing repurposing candidates through the US regulatory system.