Table 3. Selected applications of ML algorithms applied to echocardiography.

Application	ML model	Training dataset	Testing dataset	AUC	Sensitivity (%)	Specificity (%)	Accuracy (%)	Publication (Ref. #)
Differentiating between HCM & physiologic hypertrophy	SVM RF CNN	–	139 patients	0.80	96.0	77.0	–	Narula et al. (26)
Differentiating between HCM, cardiac amyloid and PAH	CNN	12,035 studies	8,666 studies	HCM 0.93, amyloid 0.87, PAH 0.85	–	–	84	Zhang et al. (27)
Differentiating between constrictive pericarditis & restrictive cardiomyopathy	SVM KL CNN RF	–	94 patients	0.96	–	–	93.7	Sengupta et al. (28)
Classifying still echo image captures into apical 2 chamber, apical 4 chamber and apical long-axis view	SVM KL RF CNN	210 clips	99 clips	–	–	–	95.0	Khamis et al. (29)
Classifying still echo image captures into 15 standard echo views	CNN	240 patients	27 patients	–	–	–	91.7	Madani et al. (30)
Classifying echo studies according to the ASE/EACVI diagnostic algorithm for diastolic dysfunction severity	CNN	6,182 studies	1,546 studies	–	–	–	99.0	Jiang et al. (31)
Assessment of myocardial velocity	KL	–	55 patients	–	73.2	78.4	–	Sanchez-Martinez et al. (32)
Detecting wall motion abnormality	CNN	–	61 patients	–	81.0	65.4	75.0	Omar et al. (33)
Quantifying MR	SVM	5,004 clips	–	–	99.4	99.6	99.5	Moghaddasi et al. (34)

ML, machine learning; CNN, convolutional neural network; AUC, area under curve; SVM, support vector machine; RF, random forests; HCM, hypertrophic cardiomyopathy; PAH, pulmonary arterial hypertension; MR, mitral regurgitation.