Predicting overall survival in synchronous metastatic nasopharyngeal carcinoma using a stacking ensemble machine learning model: a multicenter retrospective study - Report - MDSpire
Advertisement
Predicting overall survival in synchronous metastatic nasopharyngeal carcinoma using a stacking ensemble machine learning model: a multicenter retrospective study
Clinical Report: Utilizing a Stacking Ensemble Machine Learning Approach to Forecast Overall Survival in Synchronous Metastatic Nasopharyngeal Carcinoma
Overview
This study developed a stacking ensemble machine learning model to predict 3-year overall survival (OS) in patients with synchronous metastatic nasopharyngeal carcinoma (smNPC). The model demonstrated a high internal validation AUC of 0.8358.
Background
Synchronous metastatic nasopharyngeal carcinoma (smNPC) presents significant challenges in prognosis due to survival heterogeneity among patients. Accurate prediction of overall survival is crucial for treatment strategies. Current risk stratification methods are limited.
Data Highlights
Metric
Value
Internal Validation AUC
0.8358
3-Year Death Event Rate
60.0%
Training AUC
0.9150
Optimism-Corrected AUC
0.8415
Key Findings
The stacking ensemble model achieved the highest internal held-out validation AUC of 0.8358 for predicting 3-year OS.
The overall 3-year death event rate was 60.0%, with similar rates in the training and validation sets.
After multiplicity adjustment, the stacking model showed significantly higher AUC than GBDT, AdaBoost, and hard voting.
SHAP analysis identified key predictors including immunotherapy, first-line regimen, number of metastatic lesions, and number of metastatic organs.
Overfitting diagnostics indicated a training AUC of 0.9150 and an optimism gap of 0.0792.
Clinical Implications
The stacking ensemble model provides a method for predicting 3-year OS in patients with smNPC.
Conclusion
The stacking ensemble model represents a tool for individualized prognostic assessment in smNPC.
