Table 4 Detailed class-wise performance metrics compared to a stratified random baseline.
From: Multimodal machine learning for 5-year mortality prediction after percutaneous coronary intervention
Metric | Random Baseline | Our Model | Improvement (Lift) |
|---|---|---|---|
Macro F1-score | 0.497 | 0.649 | +30.6% |
Death class F1 | 0.126 | 0.385 | 3.0x (205%) |
Death class precision | 0.127 | 0.410 | 3.2x (222%) |
Specificity | 0.868 | 0.920 | - |
