Rare-Class Collapse in ECG-Based Ventricular Tachycardia and Fibrillation Detection: A Systematic Benchmark of Class-Imbalance Mitigation from Reweighting to Cascade Classification
This study demonstrates that a two-stage cascade classification approach significantly outperforms standard reweighting and cost-sensitive methods in detecting lethal ventricular arrhythmias from imbalanced ECG data, achieving a 37% reduction in missed lethal events while highlighting the critical need for specialized metrics like missed-lethal-event rate to avoid the "rare-class collapse" phenomenon.