Related Experiment Video
Updated: Aug 6, 2026
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
Development and validation of a machine learning-based early warning model for bone metastasis in newly diagnosed
Leibo Wang1,2, Wei He2, Changyong Zhao1
1Department of Urology, Affiliated Hospital of Zunyi Medical University, Zunyi, China.
Background:
Bone metastasis (BM) is common in newly diagnosed prostate cancer (PCa), particularly in patients with advanced disease at presentation. However, the indications for bone scintigraphy remain inconsistent and may lead to unnecessary imaging in low-risk patients. This study aimed to develop and validate a machine learning model for individualized prediction of BM in patients with newly diagnosed PCa.
Methods:
We retrospectively collected data from 327 patients with newly diagnosed PCa from two tertiary hospitals. Patients were randomly assigned to a training set (n = 229) and an internal validation set (n = 98). The Boruta algorithm was used to identify significant predictors. Seven machine learning models, including random forest and logistic regression, were developed and evaluated using five-fold cross-validation. Model performance was evaluated using receiver operating characteristic (ROC) curves, calibration, and decision curve analysis (DCA). The best-performing model was interpreted using SHapley Additive exPlanations (SHAP) and deployed as an online prediction tool.
Results:
Six predictors were identified by the Boruta algorithm: clinical T stage, Gleason score, total prostate-specific antigen (tPSA), alkaline phosphatase (ALP), regional lymph node metastasis, and fibrinogen. Among the seven models, the random forest model achieved the best performance, with an area under the curve (AUC) of 0.902 in the training set and 0.906 in the internal validation set. Calibration curves showed good agreement between predicted and observed outcomes, and decision curve analysis indicated favorable clinical utility. An interactive online prediction tool was developed for individualized risk estimation.
Conclusion:
We developed and internally validated an interpretable random forest model for predicting BM in newly diagnosed PCa. This model may help identify high-risk patients and guide the use of bone scintigraphy. Prospective multicenter studies with external validation are needed to further confirm its generalizability.
