Table 8. Performance comparison between our approach and existing ML-based HF mortality diagnosis methods.
| Study | Data source | Data balancing method | Multi-objective optimization | Performed model | Achieve accuracy | Explainable AI |
|---|---|---|---|---|---|---|
| Newaz, Ahmed & Haq (2021) | FIOC | SMOTE | (✗) | RF | 76.25% | (✗) |
| Ishaq et al. (2021) | FIOC | SMOTE | (✗) | ET | 92.62% | (✗) |
| Le et al. (2021) | UCI | (✗) | (✗) | RF | 74% | (✗) |
| Hussain et al. (2021) | Physionet databases | (✗) | (✗) | SVM | 88.79% | (✗) |
| Sutradhar et al. (2023b) | FIOC | SMOTE-ENN | (✗) | IBS | 92.75% | (✗) |
| Li et al. (2022) | The eICU-CRD (v-0.2) | (✗) | (✗) | XGB | 82.6% | SHAP |
| Mishra (2022) | FIOC | SMOTE | (✗) | SVM | 83.33% | (✗) |
| Plati et al. (2021) | IUHI | SMOTE | (✗) | ROT | 91.23% | (✗) |
| Nishat et al. (2022) | FIOC | SMOTE-ENN | (✗) | RF | 90% | (✗) |
| Sabahi, Vali & Shafie (2023) | PRCVD | SMOTE | (✗) | XGB | 76.4% | (✗) |
| Luo et al. (2022) | MIMIC | (✗) | (✗) | XGB | 83.1% | (✗) |
| Chicco & Jurman (2020) | FIOC | (✗) | (✗) | RF | 74% | (✗) |
| Mpanya et al. (2023) | PMRCardio Database | (✗) | (✗) | RF | 88% | (✗) |
| Mohan, Thirumalai & Srivastava (2019) | UCI | (✗) | (✗) | HRFLM | 88.7% | (✗) |
| Rahman et al. (2023) | Physionet | (✗) | (✗) | Stacking | 89.41% | (✗) |
| Sutradhar et al. (2023c) | FIOC | BOO-ST | (✗) | CBCEC | 93.67% | (✗) |
| This study | FIOC | SMOTE and Tomeklink | NSGA-II | SEHM | 94.87% | LIME |