Clinical Report: Predicting Survival Outcomes in Liver Metastases from Colorectal Cancer
Overview
This study developed and validated a machine learning model to predict long-term survival in patients with colorectal cancer liver metastasis (CRLM). The model incorporates Traditional Chinese Medicine (TCM) interventions and demonstrates robust predictive performance, offering a web-based tool for individualized risk assessment.
Background
Colorectal cancer liver metastasis (CRLM) is a significant contributor to cancer-related mortality, with a five-year survival rate of less than 30%. Accurate prognostic tools are essential for tailoring treatment strategies and improving patient outcomes. Traditional models often fail to account for complex interactions among clinical variables and therapeutic interventions.
Data Highlights
Model
AUC (36-month)
AUC (60-month)
XGBoost
0.891 (training)
0.833 (testing)
Key Findings
The optimized XGBoost model outperformed other machine learning algorithms in predicting survival.
SHAP analysis identified TNM stage, liver metastasis burden, and TCM intervention intensity as key prognostic factors.
TCM exposure showed a protective association with survival in a dose-dependent manner.
The model was validated with a temporal dataset, confirming its robustness and stability.
A web-based tool was developed for clinicians to input patient data and receive real-time risk estimates.
Clinical Implications
The machine learning model provides a novel approach to prognostic assessment in CRLM, integrating TCM interventions that may enhance survival outcomes. Clinicians can utilize this tool for personalized treatment planning and improved patient management.
Conclusion
The study successfully developed an interpretable machine learning model for predicting survival in CRLM patients, highlighting the potential benefits of integrating TCM into clinical practice. This model serves as a practical resource for enhancing individualized patient care.
