Construction and validation of a machine learning model for predicting anastomotic leak following radical esophagectomy for esophageal cancer.
Random forest model predicted post-esophagectomy anastomotic leak with strong discrimination in validation cohort
Construction and validation of a machine learning model for predicting anastomotic leak following radical esophagectomy for esophageal cancer.
Early identification of patients at high risk of anastomotic leak (anastomotic leak) following esophagectomy is essential for improving surgical outcomes.
This study aimed to predict anastomotic leak risk in the esophageal cancer (esophageal cancer) population by developing and validating a machine learning (machine learning)-based model using exclusively preoperative and baseline clinical data.
A retrospective cohort of esophageal cancer patients who underwent radical esophagectomy at the Affiliated Tumor Hospital of Xinjiang Medical University from January 2020 to May 2025 was analyzed.
correctly ranked leak risk in most cases, scale runs 0.5 to 1.0
The Random Forest model yielded a sensitivity (sensitivity) of 0.879 and a specificity (specificity) of 0.571.
SHAP analysis identified monocytes, carcinoembryonic antigen (carcinoembryonic antigen), neutrophil-to-lymphocyte ratio (neutrophil-to-lymphocyte ratio), urine creatinine (urine creatinine), and T stage as the five most influential predictors of anastomotic leak.
Calibration curves for the ensemble models demonstrated good agreement between predicted probabilities and observed outcomes.
The Random Forest model, incorporating five routinely available preoperative variables, exhibited robust discriminative performance with high sensitivity for predicting in-hospital anastomotic leak following esophagectomy.
The proposed threshold-based risk stratification approach may facilitate individualized perioperative monitoring and management.