Table 3.

Performance of the standalone AI model on 2 external validation datasets.

Hospital	Category	AUC [95% CI]	Sensitivity [95% CI]	F1 Score [95% CI]	Specificity [95% CI]	Accuracy [95% CI]	NPV [95% CI]
A	Lungs	0.886 [0.854, 0.915]	0.846 [0.788, 0.899]	0.690 [0.636, 0.740]	0.878 [0.856, 0.900]	0.873 [0.851,0.893]	0.966 [0.952, 0.978]
	Pleura	0.867 [0.812, 0.917]	0.779 [0.672, 0.873]	0.598 [0.510, 0.677]	0.936 [0.920, 0.952]	0.925 [0.907,0.941]	0.982 [0.972, 0.991]
	Cardiac	0.919 [0.846, 0.979]	0.877 [0.739, 1.000]	0.404 [0.288, 0.508]	0.925 [0.907, 0.941]	0.923 [0.905,0.940]	0.996 [0.992, 1.000]
	Aggregate	0.910 [0.883, 0.934]	0.876 [0.829, 0.918]	0.767 [0.722, 0.807]	0.885 [0.862, 0.906]	0.883 [0.862,0.902]	0.962 [0.946, 0.975]
B	Lungs	0.902 [0.835, 0.959]	0.898 [0.781, 1.000]	0.762 [0.641, 0.857]	0.873 [0.811, 0.926]	0.878 [0.823,0.924]	0.969 [0.932, 1.000]
	Pleura	0.871 [0.734, 0.989]	0.780 [0.529, 1.000]	0.650 [0.400, 0.833]	0.947 [0.907, 0.980]	0.934 [0.892,0.975]	0.980 [0.956, 1.000]
	Cardiac	0.855 [0.641, 0.996]	0.750 [0.333, 1.000]	0.488 [0.154, 0.741]	0.939 [0.899, 0.974]	0.930 [0.886,0.968]	0.988 [0.966, 1.000]
	Aggregate	0.919 [0.864, 0.968]	0.920 [0.828, 1.000]	0.820 [0.723, 0.898]	0.887 [0.828, 0.941]	0.896 [0.848,0.943]	0.970 [0.934, 1.000]
A + B (Entire dataset)	Lungs	0.889 [0.860, 0.917]	0.855 [0.802, 0.903]	0.702 [0.651, 0.746]	0.878 [0.856, 0.899]	0.874 [0.853,0.892]	0.966 [0.953, 0.978]
	Pleura	0.867 [0.818, 0.913]	0.779 [0.686, 0.869]	0.605 [0.525, 0.682]	0.938 [0.923, 0.952]	0.926 [0.911,0.941]	0.982 [0.973, 0.990]
	Cardiac	0.906 [0.836, 0.965]	0.852 [0.720, 0.967]	0.417 [0.312, 0.514]	0.927 [0.911, 0.942]	0.924 [0.909,0.940]	0.995 [0.990, 0.999]
	Aggregate	0.912 [0.888, 0.934]	0.883 [0.840, 0.920]	0.775 [0.736, 0.811]	0.885 [0.863, 0.906]	0.885 [0.865,0.903]	0.963 [0.949, 0.975]