What question did this study set out to answer?

This research aims to identify predictors of local treatment failure after stereotactic radiotherapy for brain metastases using machine learning and statistical modeling.

January 16, 2026

Lesion-Level Prediction of Local Recurrence After Stereotactic Radiotherapy for Brain Metastases Using Machine Learning and Mixed-Effects Modeling

Key Points

This research aims to identify predictors of local treatment failure after stereotactic radiotherapy for brain metastases using machine learning and statistical modeling.
Retrospective analysis of 211 brain metastases from 63 patients treated with SRT.
Features included age, lesion length, biological effective dose, sex, and Karnofsky performance score.
Random forest classifiers were applied with cross-validation to assess model performance.
Generalized linear mixed-effects modeling was utilized to confirm significant predictors.
14.7% of lesions developed local treatment failure after SRT.
The ML model achieved an AUC of 0.88 and accuracy of 0.84.
Lesion length ≥19.3 mm and biological effective dose ≤60.0 Gy were identified as critical thresholds.

Abstract

Abstract Purpose The aim of this study was to identify and validate clinically meaningful predictors of local treatment failure (LTF) after stereotactic radiotherapy (SRT) for brain metastases by integrating machine learning (ML)–based feature selection with generalized linear mixed-effects modeling (GLMM). Methods The retrospective study included 211 brain metastases from 63 patients treated with SRT. Each lesion was investigated independently, considering the within-patient clustering. Random forest classifiers were trained using a 3-fold cross-validation repeated 100 times to assess predictive performance and feature importance. The predictors included age, lesion length, biological effective dose (BED), sex, post-radiation systemic therapy, Karnofsky performance score (KPS), type of primary tumor, and lesion location. Features with high importance were further evaluated using GLMM to determine statistical significance. ROC analysis was performed to determine optimal thresholds and diagnostic accuracy. Results The median duration of follow-up was 232 days, 31 lesions (14.7%) developed LTF following SRT. The ML model achieved a mean AUC of 0.88 and an accuracy of 0.84. Variable importance rankings identified age, primary tumor, BED, length, and KPS. GLMM confirmed associations of lower BED (OR: 0.89, 95% CI: 0.80—0.98; p = 0.023) and greater lesion length (OR: 1.16, 95% CI: 1.08—1.24; p 0.001) with LTF. ROC analysis identified BED ≤60.0 Gy (AUC 0.72; sensitivity 0.81; specificity 0.71), and lesion length ≥19.3 mm (AUC 0.78; sensitivity 0.58; specificity 0.88) as thresholds. Conclusion Lesion length was the most robust predictor of LTF after SRT, with a length ≥19.3 mm indicating increased risk.

Mark Helpful

Bookmark

Relay