Scholarlink

<p>Prognostic heterogeneity remains a challenge for non-metastatic renal cell carcinoma (RCC) patients following radical nephrectomy (RN). This study aimed to develop and validate multiple machine learning models to predict overall survival (OS) in these patients using the SEER database. A retrospective analysis was conducted on patients diagnosed with non-metastatic RCC who underwent RN between 2004 and 2022, identified from the SEER database. The dataset was randomly split into training and internal validation sets at a 7:3 ratio. An external validation set included 207 eligible patients from the first Affiliated Hospital of Nanchang University and the First Hospital of Putian City. Least absolute shrinkage and selection operator (LASSO)-Cox regression and multivariate Cox proportional hazards models were used to identify independent prognostic factors for OS. Based on these variables, five machine learning models were constructed: support vector machine (SVM), k-nearest neighbors (KNN), gradient boosting decision tree (GBDT), random forest (RF), and extreme gradient boosting (XGBoost). Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), accuracy, precision, F1-score, calibration curves, and decision curve analysis (DCA). The best-performing model was interpreted using SHapley Additive exPlanations (SHAP) to identify key predictive features and their contributions. A total of 49,381 non-metastatic RCC patients were included. LASSO-Cox and multivariate Cox analyses revealed that age, sex, race, marital status, pathological grade, histologic type, T stage, N stage, radiotherapy, chemotherapy, and tumor size were significantly associated with OS. The GBDT model outperformed other models in predicting 1-, 3-, and 5-year OS, with AUCs of 0.790, 0.765, and 0.752, respectively, along with high accuracy, precision, and F1-scores. Calibration curves and DCA demonstrated favorable calibration and clinical utility. The model also showed robust performance in the external validation cohort. SHAP analysis identified age, tumor size, and T stage as the most influential predictors. We successfully developed and validated an interpretable machine learning model for predicting OS in non-metastatic RCC patients after RN. This model may assist in prognostic stratification and support individualized treatment planning.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning-based prediction of overall survival in non-metastatic renal cell carcinoma after radical nephrectomy

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月13日

查看全文
  • ChangGui Li,
  • Jie Xu,
  • Ru Chen,
  • YuanTian Lin

摘要

Prognostic heterogeneity remains a challenge for non-metastatic renal cell carcinoma (RCC) patients following radical nephrectomy (RN). This study aimed to develop and validate multiple machine learning models to predict overall survival (OS) in these patients using the SEER database. A retrospective analysis was conducted on patients diagnosed with non-metastatic RCC who underwent RN between 2004 and 2022, identified from the SEER database. The dataset was randomly split into training and internal validation sets at a 7:3 ratio. An external validation set included 207 eligible patients from the first Affiliated Hospital of Nanchang University and the First Hospital of Putian City. Least absolute shrinkage and selection operator (LASSO)-Cox regression and multivariate Cox proportional hazards models were used to identify independent prognostic factors for OS. Based on these variables, five machine learning models were constructed: support vector machine (SVM), k-nearest neighbors (KNN), gradient boosting decision tree (GBDT), random forest (RF), and extreme gradient boosting (XGBoost). Model performance was evaluated using the area under the receiver operating characteristic curve (AUC), accuracy, precision, F1-score, calibration curves, and decision curve analysis (DCA). The best-performing model was interpreted using SHapley Additive exPlanations (SHAP) to identify key predictive features and their contributions. A total of 49,381 non-metastatic RCC patients were included. LASSO-Cox and multivariate Cox analyses revealed that age, sex, race, marital status, pathological grade, histologic type, T stage, N stage, radiotherapy, chemotherapy, and tumor size were significantly associated with OS. The GBDT model outperformed other models in predicting 1-, 3-, and 5-year OS, with AUCs of 0.790, 0.765, and 0.752, respectively, along with high accuracy, precision, and F1-scores. Calibration curves and DCA demonstrated favorable calibration and clinical utility. The model also showed robust performance in the external validation cohort. SHAP analysis identified age, tumor size, and T stage as the most influential predictors. We successfully developed and validated an interpretable machine learning model for predicting OS in non-metastatic RCC patients after RN. This model may assist in prognostic stratification and support individualized treatment planning.