Table 2.
Existing literature: contributions and research gaps.
| Ref | Dataset | Methodology | Optimization Technique | Accuracy | Research Gap |
|---|---|---|---|---|---|
| 32 | UCI Heart Disease | HyOPTXGBoost and HyOPTRF | Optuna Optimization | 94.2% | Lack of interpretability |
| 33 | Cleveland | Supervised ML Algorithms | Traditional ML | 92.8% | Limited feature selection optimization |
| 34 | Framingham | Hybrid XGBoost with Optuna tuning | Hyperparameter tuning | 96.1% | High sensitivity to outliers |
| 35 | MIT-BIH Arrhythmia | ML algorithms | Feature Selection | 93.7% | Computationally expensive |
| 36 | UCI Heart Disease | Genetic Algorithm | GA | 95.3% | High training time |
| 37 | ECG-based Dataset | CNN-LSTM | CNN-LSTM | 92.4% | High computational cost |
| 38 | Custom CVD Dataset | Deep Reinforcement Learning for heart disease detection | DRL | 90.9% | Slow convergence |
| 39 | UCI Heart Disease | Gradient-Based Optimizer | GBO | 95.1% | Sensitive to parameter selection |
| 40 | ECG-based Dataset | Rime Optimization Algorithm for ECG-based detection | RIME | 91.7% | High computational cost |
| 41 | MIT-BIH Dataset | Heap-Based Optimizer (HBO) for ML-based heart disease detection | HBO | 93.2% | Lack of interpretability |