SURVIBENE-LT: A Survivorship-Based Benefit Model for Optimized Liver Transplantation Allocation - Project Abstract The only potentially life-saving recourse for end-stage liver disease is liver transplantation (LT). In the 2024, the United States witnessed 10,835 LT procedures, accounting for a staggering $8 billion in costs. Tragically, many patients succumb while on the waitlist, highlighting critical gaps in organ allocation. While the introduction of the Model for End Liver Disease (MELD), an urgency model, has been pivotal in prioritizing the sickest patients and reducing waitlist 90-day mortality, indicators of clinical trajectories, comorbidities, frailty, functional status, psychosocial risk, and social and financial support. It also fails to predict likely wait time for a matching organ. Not surprisingly, the resulting system lacks objectivity, personalization, and transparency; varies widely across patients and transplant centers – as set by the Organ for Procurement and Transplant Network (OPTN) Final Rule regulatory framework; and fails to optimize pre-LT survival and post-LT survivorship benefits. To optimize outcomes within the constraints of organ availability, we must develop objective models that minimize WL mortality, maximize post-LT benefit, and reduce unwarranted variation in care. We propose to develop novel evidence-based, personalized, clinically useful tools that minimize pre-LT mortality, maximize post-LT benefit, increase transparency, and reduce unwarranted variation in care. We have three aims: (i) Aim 1: Develop a data-driven system for waitlisting decisions and prioritizing patients for LT that combines a personalized urgency model with regionally informed predictive modeling of organ availability, (ii) Aim 2: Identify key factors influencing post-LT recipient and graft survival, complications, and functional recovery to inform personalized care planning, and (iii) Aim 3: Develop and evaluate a robust and privacy-preserving federated learning system. we will leverage a unique data infrastructure combining real- world, multimodal, longitudinal data from three geographically and demographically diverse transplant centers—University of Florida (UF), University of Minnesota (UMN), and the Cleveland Clinic Foundation (CCF)—with data from the national Scientific Registry of Transplant Recipients (SRTR), capturing ~30K patients referred for transplant evaluations, where around 10K. were waitlisted. By combining real-world data with stakeholder-driven outputs, SURVIBENE-LT will enable earlier identification of the highest-risk patients, optimize post-LT benefit, reduce unwarranted variation in care, and enhance patient-centric decision making. Our discovery of novel patient, donor, and perioperative contributors to survivorship will support development and testing of clinical interventions to improve patient outcomes across the care continuum. This work directly addresses NIH-NIDDK priorities by advancing personalized, evidence-based clinical decision making, outcome optimization, and a framework for generalizability not seen before in LT care.
