Table 2 Literature review comparison table.
From: Predicting cardiovascular risk with hybrid ensemble learning and explainable AI
References | Methods used | Limitations | Novelty of the approach | Results achieved |
|---|---|---|---|---|
KNN, SVM, Logistic Regression, Random Forest | Imbalanced dataset, Limited accuracy of base models | Comparison of multiple classifiers and use of SMOTE for data imbalance | Random Forest performed best with improved accuracy using SMOTE | |
SVM, KNN, Naïve Bayes | Limited dataset, Imbalanced data | Comparison of various ML classifiers | Random Forest performed best with the highest accuracy for coronary artery prediction | |
Random Forest, KNN, LR, SVM | Imbalanced data, Overfitting | Application of a novel Grey Wolf Algorithm for feature selection | These models achieved good prediction accuracy for Coronary Heart Disease Classification | |
XGBoost, Random Forest, Artificial Neural Networks | High computational cost, Overfitting | Proposed SHAP analysis to interpret model predictions | Achieved high predictive performance with SHAP interpretability | |
Random Forest, XGBoost, Decision Tree, K-Means, Fuzzy C-Means | Data imbalance, High computational cost | Stacked ensemble learning for heart failure survival prediction | Good accuracy, precision, recall, and F1 score | |
Naive Bayes, Neural Networks, Decision Trees | Dataset imbalance, Small sample size | Focused on early detection using hybrid models | Logistic Regression showed highest performance on processed datasets | |
CatBoost, XGBoost, LightGBM, Random Forest, Neural Networks | Dataset imbalance, Computation time | Hybrid model combining various classifiers for improved accuracy | Achieved an AUC-ROC of 0.82 for cardiovascular risk prediction | |
Support Vector Machines, K-Nearest Neighbors, Gradient Boosting | Data preprocessing challenges, Imbalanced data | Discussed hybrid approach with data imbalance correction via SMOTE | Voting Ensemble achieved improved model accuracy |
