loading
loading
NIH grant 1R01DK148650-01: A Temporal Causal Knowledge Base of Liver Disease Progression in Liver Transplant Candidates
Institute: {'code': 'DK', 'abbreviation': 'NIDDK', 'name': 'National Institute of Diabetes and Digestive and Kidney Diseases', 'admin_org_id': '12500020', 'admin_funding_url': ''}. PI: Lisiane Pruinelli. Award: $780358.
Start: 2026-09-18T00:00:00.
Project Abstract
Liver transplantation (LT) is the only life-saving treatment for end-stage liver disease (ESLD), yet demand for
donated organs far exceeds supply. More than 11,000 patients are added to the waitlist for LT in the U.S. each
year, 20% of whom die prior to being matched with an organ or get delisted because they become too sick to
undergo the procedure. Among patients who undergo LT, outcomes (e.g., survival, hospital readmission) vary
considerably. Patient history and the quality and timing of care for waitlisted patients influence the success of
LT and subsequent survival and care utilization. However, a critical gap in knowledge regarding the sequence,
determinants, and timing of LD progression has hindered our ability to optimize care. LD progression is
characterized by a vicious cycle in which the processes of hepatic decompensation (HD)—the within-liver
cascade from cirrhosis to loss of synthetic/clearance function with events such as ascites, visceral bleeding,
jaundice—lead to worsening of preexisting or new-onset extra-hepatic comorbidities/complications (EHC; e.g.,
infection, renal dysfunction, frailty), a “domino” effect. Current LT research and practice often lack a temporal
dimension, focusing on a single snapshot of patient status and largely disregarding the cumulative impact of
HD and EHC. We propose to address this gap by constructing a causal-temporal knowledge base that
informs patient management and, in the long term, organ allocation for transplantation. To date, the absence of
comprehensive temporal clinical data has hindered the creation of such a knowledge base. We are proposing
a multicenter study in which we link more than 10 years of longitudinal data on ~10,000 waitlisted patients from
the electronic health records (EHRs) of four geographically distinct transplant centers with data from the gold-
standard national transplant registry. Through innovative application of causal and temporal modeling
methods to this comprehensive dataset, which includes all data points and time stamps (e.g., diagnoses,
laboratories), we will develop essential temporal and causal data analysis techniques. Specifically, we will 1)
reconstruct the population-based timeline of LD progression, developing metrics for ESLD severity, changes in
severity over time, risk of HD and EHC, and prognostic information; 2) create an individual-level predictive
model of LD progression, enabling estimation of the rate of disease progression, time to LT, and personalized
risk of comorbidities/complications at any time t before LT; and 3) establish causal relationships among EHCs,
markers of HD, and outcomes, identifying modifiable causal factors as intervention targets. Our
interdisciplinary team comprises transplant surgeons, hepatologists, machine learning experts, causal
inference specialists, and policy researchers. With our experience in LT research, disease trajectory analysis,
and causal discovery from observational data, we are poised to successfully execute this transformative
research, significantly advancing efforts to optimize care of waitlisted LT patients. Our methods and results will
support clinical trials of LD interventions, expanded models, and application to other chronic diseases.document fetched 2026-09-27 02:45