Background <p>Accurate mortality prediction is vital for liver transplant allocation, but models like MELD-Na and Child–Pugh may overlook important individual risk factors. This study aimed to develop and validate a prognostic model using key clinical variables and both statistical and machine learning methods.</p> Materials and methods <p>This retrospective study of 193 liver transplant candidates at Taleghani Hospital analyzed clinical data to identify predictors of waiting list mortality. Key variables were selected based on univariate logistic regression performance and clinical relevance, avoiding overlap with composite scores. These predictors were then modeled using logistic regression and imbalance-handled machine learning approaches, and a nomogram was constructed for individualized risk estimation.</p> Results <p>Among the 193 patients included in the study, 9 individuals (4.7%) died while on the liver transplant waiting list. In the multivariate logistic regression analysis, both the MELD-Na score (odds ratio [OR] = 1.25, 95% confidence interval [CI]: 1.06–1.47) and hepatic encephalopathy (OR = 36.94, 95% CI: 5.07–269.28) remained statistically significant. MELD-Na (AUC: 0.86), albumin (AUC: 0.77), and hepatic encephalopathy (AUC: 0.62) were retained as the key non-overlapping predictors of mortality. Among the imbalance-handled models, support vector machine (AUC: 0.85; PPV: 0.41) and naïve Bayes (AUC: 0.88; PPV: 0.38) demonstrated the highest overall performance. Additionally, a clinically interpretable nomogram constructed from the key predictors showed promising performance.</p> Conclusion <p>This study developed internally validated prognostic models and a clinically interpretable nomogram to predict mortality among liver transplant candidates. The findings support individualized risk assessment and represent an important first step toward integrating machine learning into transplant decision-making. However, external and multicenter validation is required before clinical implementation.</p>