Machine Learning-Based Prediction of Local Recurrence in Uveal Melanoma After Ruthenium-106 Plaque Brachytherapy...

Area of Science:

• Ophthalmology
• Oncology
• Medical Informatics

Background:

• Uveal melanoma (UM) is the most common adult intraocular malignancy.
• Local recurrence risk after plaque brachytherapy depends on treatment and follow-up.
• Predicting UM recurrence is crucial for effective patient management.

Purpose of the Study:

• To evaluate machine learning (ML) models for predicting UM local recurrence.
• To integrate demographic and clinical data for UM recurrence prediction.
• To assess the effectiveness of Ruthenium-106 plaque brachytherapy outcomes.

Main Methods:

• Retrospective analysis of 167 UM patients treated with Ru-106 plaque brachytherapy.
• Feature selection using recursive feature elimination (RFE) and dimensionality reduction with principal component analysis (PCA).
• Training and evaluation of six ML classifiers, including random forest (RF).

Main Results:

• The RF model achieved an AUC of 0.86, accuracy of 0.82, and recall of 0.91.
• Key predictors identified by SHAP analysis included largest basal diameter, plaque size, thickness, apex dose, diabetes mellitus, and transpupillary thermotherapy.
• The RF model demonstrated superior performance in predicting UM local recurrence.

Conclusions:

• ML models, especially RF, accurately predict UM local recurrence.
• These predictive models can enhance personalized treatment strategies for UM patients.
• Further validation in larger, multicenter studies is recommended to refine predictive accuracy.