RETRACTED: Machine-Learning-Based Survival Prediction in Castration-Resistant Prostate Cancer: A Multi-Model Analysis...

Area of Science:

• Oncology
• Medical Informatics
• Biostatistics

Background:

• Accurate survival prediction is crucial for treatment planning in castration-resistant prostate cancer (CRPC).
• Traditional statistical models have limitations in handling complex data interactions and variable inclusion for CRPC prognosis.
• Machine learning (ML) offers potential for more robust survival prediction in CRPC.

Purpose of the Study:

• To develop and evaluate ML models for predicting cancer-specific mortality (CSM), overall mortality (OM), and short-term survival in CRPC patients.
• To compare the performance of ML models against traditional statistical methods for CRPC survival prediction.
• To identify key predictors of survival outcomes in CRPC using interpretable ML techniques.

Main Methods:

• Retrospective collection of 46 variables from 801 CRPC patients.
• Development of multiple ML models (RSF, XGBoost, LightGBM, logistic regression) for survival prediction.
• Performance evaluation using C-index, AUC, accuracy, precision, recall, F1-score, and SHAP for interpretability.

Main Results:

• RSF models achieved the highest C-index for CSM (0.772) and OM (0.771).
• RSF and XGBoost models showed superior performance in predicting 2- and 3-year survival, respectively.
• SHAP analysis identified time to first-line CRPC treatment, hemoglobin, and alkaline phosphatase as key prognostic factors.

Conclusions:

• ML models, particularly RSF and XGBoost, outperform traditional methods in predicting CRPC survival.
• These ML models provide accurate and interpretable prognostic tools for personalized CRPC treatment.
• External validation and incorporation of novel therapies are recommended for clinical implementation.