Table 3.
Summarized studies and used algorithms.
| Author | Study | Algorithm | Performance (Accuracy/AUROC/Precision) |
|---|---|---|---|
| Kogan et al. [49] | A machine learning approach to identifying patients with pulmonary hypertension using real-world electronic health records | XGBoost | 0.92 AUROC |
| Mohammad et al. [48] | Development and validation of an artificial neural network algorithm to predict mortality and admission to hospital for heart failure after myocardial infarction: a nationwide population-based study | ANN | 0.85–0.78 AUROC |
| Moghaddasi et al. [37] | Automatic assessment of mitral regurgitation severity based on extensive textural features on 2D echocardiography videos | SVM | 99.45% |
| Attia et al. [38] | Screening for cardiac contractile dysfunction using an artificial intelligence-enabled electrocardiogram | CNN | 85.70% |
| Porumb et al. [39] | Precision medicine and artificial intelligence: a pilot study on deep learning for hypoglycemic events detection based on ECG | CNN, RNN, Grad-CAM | 82.40–85.70% |
| Salte et al. [40] | Artificial intelligence for automatic measurement of left ventricular strain in echocardiography | ANN | 97–98% |
| Kusunose et al. [41] | A deep learning approach for assessment of regional wall motion abnormality from echocardiographic images | CNN | 0.99–0.97 AUROC |
| Berikol et al. [45] | Diagnosis of acute coronary syndrome with a support vector machine | SVM, ANN, NB, Logistic Regression | 90.10–99.13% |
| Motwani et al. [46] | Machine learning for prediction of all-cause mortality in patients with suspected coronary artery disease: a 5-year multicentre prospective registry analysis | DT—LogitBoost | 0.79 AUROC |
| Kakadiaris et al. [15] | Machine learning outperforms ACC/AHA CVD risk calculator in MESA | SVM | 0.92 AUROC |
| Kanwar et al. [50] | Risk stratification in pulmonary arterial hypertension using Bayesian analysis | Bayesian Network | 0.80 AUROC |
| Galloway et al. [43] | Development and validation of a deep-learning model to screen for hyperkalemia from the electrocardiogram | CNN | 0.85–0.88 AUROC |
| Luongo et al. [44] | Machine learning enables noninvasive prediction of atrial fibrillation driver location and acute pulmonary vein ablation success using the 12-lead ECG | DT | 78.26% |
| Karwath et al. [55] | Redefining β-blocker response in heart failure patients with sinus rhythm and atrial fibrillation: a machine learning cluster analysis | Hierarchical clustering, Variational autoencoders (VAEs), K-means | N/A |
| Cikes et al. [56] | Machine learning-based phenogrouping in heart failure to identify responders to cardiac resynchronization therapy | K-means, Multiple Kernel Learning | N/A |
| Li et al. [59] | Identification of type 2 diabetes subgroups through topological analysis of patient similarity | Topological data analysis | N/A |
| Ghesu et al. [61] | Multi-scale deep reinforcement learning for real-time 3D-landmark detection in CT scans | Deep reinforcement learning | Accuracy improved by 20–30% |
| Levyid et al. [60] | Applications of machine learning in decision analysis for dose management for dofetilide | PCA, K-means, reinforcement learning—SARSA | 86–93% |
| Garvin et al. [62] | Automating quality measures for heart failure using natural language processing: a descriptive study in the department of veterans affairs | NLP | Precision 98.7% |
| Shah et al. [63] | Impact of different electronic cohort definitions to identify patients with atrial fibrillation from the electronic medical record | NLP | 0.89 AUROC |
| Kaspar et al. [64] | Underestimated prevalence of heart failure in hospital inpatients: a comparison of ICD codes and discharge letter information | NLP | Precision 96% |
| Patel et al. [65] | Development and validation of a heart failure with preserved ejection fraction cohort using electronic medical records | NLP | Precision 96% |
| Mahajan et al. [66] | Combining structured and unstructured data for predicting risk of readmission for heart failure patients | NLP | 0.65 AUROC |
| Galper et al. [67] | Comparison of adverse event and device problem rates for transcatheter aortic valve replacement and mitraclip procedures as reported by the transcatheter valve therapy registry and the Food and Drug Administration postmarket surveillance data | NLP | N/A |
| Afzal et al. [69] | Mining peripheral arterial disease cases from narrative clinical notes using natural language processing | NLP | 91.8% |
| Ashburner et al. [70] | Natural language processing to improve prediction of incident atrial fibrillation using electronic health records | NLP | N/A |