A Retrospective Cohort Study on Predicting Delayed Graft Function in Liver Transplant Patients with Hepatocellular Carcinoma: a Nomogram and Machine Learning Approaches.

A Retrospective Cohort Study on Predicting Delayed Graft Function in Liver Transplant Patients with Hepatocellular Carcinoma: a Nomogram and Machine Learning Approaches.

Background: This study aimed to develop a predictive model for delayed graft function (DGF) in liver transplant patients with hepatocellular carcinoma (HCC) based on preoperative biochemical indicators, using both logistic regression and XGBoost machine learning algorithms.

Methods: A retrospective cohort study was conducted, including 131 liver transplant patients from January 2020 to April 2022. Preoperative biochemical markers and hematological parameters were analyzed. Logistic regression and XGBoost models were constructed to predict DGF, and their performance was evaluated using the area under the ROC curve (AUC). Shapley Additive Explanations (SHAP) analysis was employed to interpret the feature contributions.

Inclusion Criteria:

1. Patients with liver cancer aged between 18 and 75 years.
2. Patients who have received liver transplantation.
3. Complete preoperative and postoperative medical imaging data, clinical records and pathological test reports are available.

Exclusion Criteria:

1. Primary non-hepatocellular carcinoma
2. Recipients were younger than 18 years and older than 75 years
3. Due to incomplete or missing clinical data, this dataset did not meet the criteria for inclusion in the statistical analysis.
4. Serious complications associated with liver transplantation, such as acute transplant rejection or transplant liver failure, occur within 7 days after surgery.
5. Re-surgery within 7 days after surgery due to liver transplant-related complications or other surgical procedures, such as transplant site infection or other major surgery.