Predicting overall survival in synchronous metastatic nasopharyngeal carcinoma using a stacking ensemble machine learning model: a multicenter retrospective study - Scorecard - MDSpire
Advertisement
Predicting overall survival in synchronous metastatic nasopharyngeal carcinoma using a stacking ensemble machine learning model: a multicenter retrospective study
Clinical Scorecard: Utilizing a Stacking Ensemble Machine Learning Approach to Forecast Overall Survival in Synchronous Metastatic Nasopharyngeal Carcinoma: Findings from a Multicenter Retrospective Analysis
At a Glance
Category
Detail
Condition
Synchronous Metastatic Nasopharyngeal Carcinoma
Key Mechanisms
Stacking ensemble machine learning model for predicting 3-year overall survival.
Target Population
Patients with synchronous metastatic nasopharyngeal carcinoma (smNPC).
Care Setting
Multicenter retrospective study.
Key Highlights
The stacking ensemble achieved an AUC of 0.8358 for 3-year OS prediction.
3-year death event rate was 60.0% overall.
SHAP identified key predictors including immunotherapy and number of metastatic lesions.
Guideline-Based Recommendations
Diagnosis
Histopathologically confirmed nasopharyngeal carcinoma with distant metastasis at initial diagnosis.
Management
Consider multimodality treatment including first-line immunochemotherapy and radiotherapy.
Monitoring & Follow-up
Assess 3-year overall survival as a clinically meaningful endpoint.
Risks
Survival heterogeneity influenced by demographic characteristics and tumor burden.
Patient & Prescribing Data
413 patients with smNPC from three institutions.
Patients received multimodality treatment between January 2010 and January 2023.
Clinical Best Practices
Utilize ensemble learning methods to improve prognostic assessments.
