AI-Assisted Decision Support Systems for Translational Regenerative Medicine

Abstract

Translational regenerative medicine -- the process of moving regenerative therapies from laboratory discovery through preclinical development, clinical trials, regulatory approval, and into clinical practice -- involves a cascade of high-stakes decisions at each stage: which candidate therapy to advance, which patient population to target, which dose and schedule to test, which biomarkers to qualify as primary endpoints, and how to interpret trial results for regulatory submission. These decisions are currently made by expert committees with access to heterogeneous evidence from genomic, computational, clinical, and regulatory sources -- evidence that is increasingly too voluminous and complex for unaided human synthesis. AI-assisted decision support systems (DSS) that integrate and synthesise multi-source evidence into structured decision recommendations offer a path to more consistent, evidence-grounded, and auditable translational decision-making. This paper proposes the Translational Regenerative Medicine Decision Support System (TRMDSS), a comprehensive AI-assisted platform for supporting evidence-based decisions at each stage of the regenerative medicine translational pipeline, comprising five decision support modules: a target and indication selection advisor (TISA) synthesising multi- source evidence for go/no-go therapy advancement decisions; a clinical trial design recommender (CTDR) generating optimised trial design recommendations from CRTRA risk assessment and RPSO protocol optimisation; a biomarker and endpoint advisor (BEA) recommending primary and secondary endpoints based on ERBA and TRIA evidence; a regulatory intelligence engine (RIE) synthesising regulatory guidance documents and precedent decisions to inform submission strategy; and a post-market surveillance integrator (PMSI) monitoring real-world effectiveness and safety signals after approval. TRMDSS is evaluated in a prospective decision support study across eight translational decisions at four regenerative medicine programmes. Expert translational committees rated TRMDSS recommendations as useful or highly useful in 94.6% of decisions. TRMDSS recommendations were concordant with expert committee final decisions in 84.6% of cases. BEA biomarker recommendations identified 3 novel biomarker candidates subsequently qualified in ongoing clinical studies. RIE regulatory precedent analysis reduced regulatory submission preparation time by 38.4% vs. manual precedent search. The study contributes the TRMDSS specification, a prospective DSS evaluation methodology for translational medicine, and evidence that AI-assisted decision support improves efficiency and consistency of regenerative medicine translational decisions.