Machine learning-driven prognostication in liver transplantation: A tacrolimus intrapatient variability enhanced...

Area of Science:

• Hepatology and Transplant Surgery
• Clinical Prognostics
• Immunosuppression Management

Background:

• Accurate prediction of long-term survival post-liver transplantation (LT) is clinically challenging.
• Tacrolimus intrapatient variability (Tac-IPV) shows prognostic potential but is underutilized.
• Current prognostic markers lack sufficient accuracy for long-term outcomes.

Purpose of the Study:

• To develop and validate a novel composite index for predicting long-term survival after LT.
• To integrate Tac-IPV with other clinical factors into a predictive model.
• To compare the prognostic performance of the new index against established scoring systems.

Main Methods:

• Retrospective analysis of 381 adult LT recipients from two centers.
• Development of the Liver Transplantation Prognosis Predictor (LTPP) integrating Tac-IPV, total bilirubin, and donor age.
• Evaluation using survival analysis, random forest modeling, and external validation.

Main Results:

• The random forest model showed robust predictive performance for 1-, 2-, and 3-year survival (AUCs ranging from 0.68 to 0.76).
• LTPP significantly outperformed MELD and Child-Pugh scores in risk stratification (log-rank P < 0.0001).
• LTPP demonstrated reliable prognostic accuracy in both internal and external validation datasets.

Conclusions:

• The LTPP is a novel, clinically accessible composite index incorporating Tac-IPV.
• LTPP offers superior prognostic accuracy for long-term survival post-LT compared to existing scores.
• LTPP represents a promising tool for individualized post-transplant management and patient care.