Gene selection is essential for improving classification performance and interpretability in high-dimensional microarray data. This study applies a Multiple Support Vector Machine–Recursive Feature Elimination (MSVM-RFE) framework for gene selection in Type 2 Diabetes Mellitus (T2DM). Experiments were conducted on a GEO microarray dataset comprising 118 samples (73 controls and 45 T2DM cases) with 25,770 genes. MSVM-RFE employs multiple linear SVM models within a 10-fold cross-validation scheme as feature selection to enhance accuracy and was evaluated under different train–test splits, with and without SMOTE resampling. The selected gene subsets were classified using SVM with linear, RBF, and polynomial kernels. The best configuration achieved 95.67% accuracy, with high sensitivity, specificity, and AUROC, using fewer than 100 genes. These results demonstrate that MSVM-RFE provides a robust and effective gene selection strategy for T2DM microarray analysis.
Basir et al. (Wed,) studied this question.