Prediction of Distant Metastasis in Renal Cell Carcinoma Using Machine Learning Algorithms: A Multicenter Cohort...

Area of Science:

• Oncology
• Medical Informatics
• Machine Learning

Background:

• Distant metastasis prediction in renal cell carcinoma (RCC) remains a challenge.
• Limited machine learning (ML) studies have explored predictive models for RCC metastasis.

Purpose of the Study:

• To develop and validate ML models for predicting distant metastasis in RCC patients.
• To compare the performance of various ML algorithms using diverse data handling techniques.

Main Methods:

• Retrospective analysis of RCC data from SEER (2004-2015) and CNCC (2010-2020) databases.
• Application of seven ML algorithms with fivefold cross-validation.
• Analysis using Python on incomplete, complete, upsampled, and downsampled datasets.

Main Results:

• Neural network model performed best on incomplete data (AUC 0.7467±0.0573).
• Support Vector Machine (SVM) showed highest accuracy on complete data (AUC 0.8221±0.0485).
• Extreme Gradient Boosting (XGBoost) and SVM models demonstrated strong performance on upsampled (AUC 0.8162±0.0558) and downsampled (AUC 0.8274±0.0546) data, respectively.

Conclusions:

• ML algorithms can effectively predict distant metastasis in RCC patients.
• Optimized ML models show significant potential for clinical application in RCC management.