Development and validation of a personalized web-based calculator of aggressive recurrence after surgery for early-stage hepatocellular carcinoma by machine learning.
Journal:
Clinical & translational oncology : official publication of the Federation of Spanish Oncology Societies and of the National Cancer Institute of Mexico
Published Date:
Nov 24, 2025
Abstract
BACKGROUND: Aggressive recurrence (AR) is an important factor affecting prognosis after surgery for hepatocellular carcinoma (HCC). This study aimed to establish and evaluate a visual calculator to predict AR using by machine learning (ML) model. METHODS: Patients diagnosed with HCC at an early stage were reviewed. The prediction ability of each model was evaluated using accuracy, sensitivity, specificity, precision, F1 score, and the area under the curve (AUC). Then, the model's prediction performance was evaluated by calibration curves, decision curve analysis (DCA), and precision-recall curves (PRC). RESULTS: 483 patients were ultimately included in this study. The baseline characteristics indicate that patients in the AR group exhibit poorer liver function and more advanced tumor features. Then, nine risk features were identified and incorporated into the nine development models, respectively. Among the models, the XGBoost model showed the best prediction ability (AUC 0.986, 95% CI: 0.983-0.988). Calibration curves, DCA, and PRC further demonstrated the robust performance and clinical applicability. Then, A web-based calculator was built. CONCLUSION: An explainable XGBoost model to predict AR for patients with early-stage HCC after surgery was feasible and effective, suggesting its superior potential in tailoring surgical strategies and optimizing personalized postoperative treatment plans.
Authors
Keywords
No keywords available for this article.
