Development and Validation of a Machine Learning Model for Predicting 3-Year Overall Survival After Transjugular...

Area of Science:

• Hepatology
• Medical Informatics
• Machine Learning in Medicine

Background:

• Predicting overall survival (OS) in cirrhotic patients undergoing transjugular intrahepatic portosystemic shunt (TIPS) is complex.
• Interdependencies of clinical variables pose challenges for accurate OS prediction.

Purpose of the Study:

• Develop and validate a machine learning (ML)-based predictive model for OS in cirrhotic patients undergoing TIPS.
• Utilize preprocedural clinical variables to enhance OS prediction accuracy.

Main Methods:

• A multicenter, retrospective study of 347 cirrhotic patients undergoing TIPS.
• Trained and validated ML models (including GBM and RSF) using demographic, biochemical, and procedural data.
• Evaluated model performance using ROC curves, AUC, C-index, and Kaplan-Meier analysis.

Main Results:

• Gradient Boosting Machine (GBM) demonstrated superior performance in predicting 3-year OS.
• Key prognostic factors identified: cirrhosis etiology, hemoglobin, creatinine, and prothrombin time.
• The GBM model effectively stratified patients into high- and low-risk groups (P<0.001).

Conclusions:

• The GBM-based model offers improved OS prediction for cirrhotic patients undergoing TIPS.
• This model facilitates enhanced risk stratification and personalized treatment strategies.
• Machine learning application shows promise in managing cirrhotic patients undergoing TIPS.