Table 17.
Detecting coronary artery disease and coronary heart disease using machine learning techniques.
| Data Set/Study | Machine Learning Approach | Results |
|---|---|---|
| ——— [195] | Logistic regression, Elastic Net, SVM, Random Forest, XGBoost |
AUC CAD 1 classification LR: 0.79 ± 0.03, EN: 0.90 ± 0.03, SVM: 0.82 ± 0.03, RF: 0.83 ± 0.03, XGBoost: 0.85 ± 0.03 AUC LVEDP 2 classification LR: 0.77 ± 0.05, EN: 0.89 ± 0.03, SVM: 0.76 ± 0.04, RF: 0.73 ± 0.05, XGBoost: 0.81 ± 0.04, Sensitivity CAD EN: 80%, Specificity CAD EN: 80%, Sensitivity LVEDP EN: 91%, Specificity LVEDP EN: 81% |
| Framingham Heart Study [196] | Multilayer Perceptron | Accuracy: 96.50%, Sensitivity: 91.90%, Specificity: 98.28% |
| Cardiovascular Disease Dataset [197] | k-NN, Bagging, Binary Logistic Classification, Naive Bayes, Boosting |
Bagged Decision Accuracy: 73.9% Gradient Boosting Recall: 73.39% Neural Network F1-score: 72% XGB AUC: 73% AdaBoost Precision: 77.8% |
| MIMIC-II [198] | Random Forest | Accuracy: 96% Sensitivity: 100% Specificity: 85% |
1 Coronary artery disease; 2 Left ventricular end-diastolic pressure.