Table 3.
Performance metrics for machine learning algorithms and conventional risk scores. IV—internal validation, EV—external validation, ML—machine learning, NLP—natural language processing, HF—heart failure, HFrEF—heart failure with reduced ejection fraction, HFmrEF—heart failure with mid-range ejection fraction, HFpEF—heart failure with preserved ejection fraction HD—heart disease, CAD—coronary artery disease.
| No. | Author | Algorithm | AUC for ML in IV | AUC for MAGGIC in IV | AUC for GWTG-HF in IV | AUC for ML in EV |
AUC for MAGGIC in EV | AUC for GWTG-HF in EV |
|---|---|---|---|---|---|---|---|---|
| 1. | C. Luo et al. [33] | XGBoost | 0.831 | - | 0.667 | 0.809 | - | - |
| 2. | E. Adler et al. [34] | boosted decision tree | 0.88 | - | 0.74 | 0.81–0.84 | - | 0.758 |
| 3. | J. Kwon et al. [35] | deep neural network | - | - | - | 0.88—in-hospital mortality 0.782—12-month mortality 0.813—36-month mortality |
0.718—12-month mortality 0.729—36-month mortality |
0.728—in-hospital mortality |
| 4. | L. Jing et al. [36] | XGBoost | 0.77 | - | - | 0.78 | - | - |
| 5. | J. Chirinos [37] | created by the tree-based pipeline optimizer platform | 0.743 (C-index) |
0.621 (C-Index) |
- | 0.717 (C-index) |
0.622 (C-index) |
- |
| 6. | J. Kwon et al. [38] | deep neural network | 0.912—(HD) | - | - | 0.913 (HF) 0.898 (HD) 0.958 (CAD) |
0.806 (HF) | 0.783 (HF) |
| 7. | S. Mahajan [39] | ensemble ML | - | - | - | 0.6987 | - | - |
| 8. | S. Mahajan [40] | created by NLP process | - | - | - | 0.6494 | - | - |
| 9. | S. Kakarmath et al. [41] | The protocol for the study | ||||||