Table 2.

Classification studies performing external validation

No.	First author	Title	Type of external testing	Number of patients	Results
1	Singh	Direct Risk Assessment From Myocardial Perfusion Imaging Using Explainable Deep Learning	External testing group	20,401 patients in the training and internal testing group (5 sites) and 9,019 in the external testing group (2 different sites)	AUC: 0.73
2	Betancur	Deep Learning for Prediction of Obstructive Disease From Fast Myocardial Perfusion SPECT A Multicenter Study	Multicenter	Total of 1,638 patients (67% men) without known coronary artery disease, undergoing stress 99mTc-sestamibi or tetrofosmin MPI with new generation solid-state scanners in 9 different sites	Per patient AUC: 0.80 Per vessel: AUC: 0.76
3	Nakajima	Diagnostic accuracy of an artificial neural network compared with statistical quantitation of myocardial perfusion images: a Japanese multicenter study	Multicenter	364 patients collected from nine hospitals served as the validation dataset	Stress defects AUC: 0.92 Stress-induced ischemia AUC: 0.90 for patients with old myocardial infarction based on rest defects AUC: 0.97
4	Otaki	Clinical Deployment of Explainable Artificial Intelligence of SPECT for Diagnosis of Coronary Artery Disease	Multicenter	External testing was performed in 555 patients from 2 centers	AUC: 0.83
5	Apostolopoulos	Multi-input deep learning approach for Cardiovascular Disease diagnosis using Myocardial Perfusion Imaging and clinical data	Image acquisition device variation	98 patients	Accuracy: 79.16%
6	Hu	Machine learning predicts per-vessel early coronary revascularization after fast myocardial perfusion SPECT: results from multicentre REFINE SPECTregistry	Multicenter	1980 patients from 9 centres	Per-vessel AUC: 0.79 Per-patient AUC: 0.81
7	Betancur	Deep Learning Analysis of Upright-Supine High-Efficiency SPECT Myocardial Perfusion Imaging for Prediction of Obstructive Coronary Artery Disease: A Multicenter Study	Division of dataset in 4 groups and performed a leave one-center-out cross-validation for each center. Overall predictions for each center were merged to have an overall estimation of the multicenter performance	1160 patients	Per-patient AUC: 0.81 Per-vessel AUC: 0.77
8	Otaki	Diagnostic Accuracy of Deep Learning for Myocardial Perfusion Imaging in Men and Women with a High-Efficiency Parallel-Hole-Collimated Cadmium-Zinc-Telluride Camera: multicenter study	Training and testing datasets included both men and women for prediction of obstructive CAD using repeated leave-one-center-out external validation (4 models built from 3 centers and tested in 4th center)	1160 patients in 4 separate centers	Sensitivity: 82% in men, and 71% in women