Predicting Biochemical Recurrence After Robot-Assisted Prostatectomy with Interpretable Machine Learning Model.
Tianwei Zhang1, Hisamitsu Ide1,2, Jun Lu1
1Department of Urology, Graduate School of Medicine, Juntendo University, Tokyo 113-8421, Japan.
Journal of Clinical Medicine
|October 16, 2025
Summary
Machine learning models can predict biochemical recurrence after robot-assisted radical prostatectomy. The LightGBM model showed strong performance, identifying key predictors like pathological T stage and positive surgical margins.
Area of Science:
- Urology
- Machine Learning
- Oncology
Background:
- Biochemical recurrence (BCR) is a significant concern after robot-assisted radical prostatectomy (RARP).
- Predicting BCR aids in personalized treatment strategies and patient management.
Purpose of the Study:
- To develop and evaluate machine learning (ML) models for predicting BCR post-RARP.
- To identify key clinical features contributing to BCR prediction.
Main Methods:
- Retrospective analysis of 1125 patients undergoing RARP.
- Development and validation of five ML models using stratified sampling.
- Performance evaluation using AUC, accuracy, sensitivity, specificity, and F1 scores; interpretability via SHAP values.
Main Results:
- The LightGBM model demonstrated the highest predictive ability with an AUC of 0.881.
- Key predictors identified include pathological T stage, positive surgical margin, PSA nadir, initial PSA, biopsy positivity, seminal vesicle invasion, Grade Group, and perineural invasion.
- SHAP values provided insights into feature contributions.
Conclusions:
- Validated ML models effectively predict BCR following RARP.
- The LightGBM model, utilizing eight key variables, shows promising performance and clinical applicability for BCR prediction.
