Genetic Algorithm-Optimized support vector machine for MRI-based multiple sclerosis diagnosis from white matter lesion features.

This study evaluates a genetic algorithm-assisted support vector machine (GA-SVM) framework for multiple sclerosis (MS) diagnosis using a publicly available dataset (n=709; 370 MS, 339 controls) comprising five pre-extracted white matter lesion features. Model performance was assessed using repeated nested stratified 10×5 cross-validation, with hyperparameter optimization confined to inner training folds and results reported from aggregated out-of-fold predictions. The tuned RBF-SVM achieved AUC=0.993, accuracy=0.987, sensitivity=1.000, specificity=0.973, precision=0.976, and F1-score=0.988 at a probability threshold of 0.5. Precision-recall (AP≈0.99) and calibration analysis (Brier=0.012) confirmed strong discrimination and reliable probability estimates. Comparative benchmarking showed competitive performance while preserving model simplicity and interpretability. These findings support the feasibility of compact radiomics combined with rigorously validated classical machine learning for MS classification. External multi-site validation is required to confirm generalizability.