Table 3. Hold-out validation results (75% Train - 25% Test Split).

Metric	Mamba	Transformer	Stacking	Voting	XGBoost
Accuracy	0.9514	0.9226	0.935	0.9211	0.8954
F1 score	0.9511	0.9207	0.933	0.918	0.8915
AUC	0.9922	0.9771	0.985	0.9758	0.9605
Jaccard	0.9052	0.853	0.875	0.848	0.804
Cohen’s Kappa	0.9028	0.8452	0.87	0.842	0.7906
Sensitivity	0.9499	0.9202	0.931	0.9043	0.8802
Specificity	0.9528	0.925	0.938	0.9371	0.9099