Table 4. Comparison of Performance Metrics of KL Grading of AI Models.

	AUC	Kappa	Accuracy	Sensitivity	Specificity	Test
Tiulpin et al. (47)	0.93					External (OAI)
Liu et al. (48)			0.83	0.78	0.95	Internal
Norman et al. (49)				KL0, 1 = 71%	KL0, 1 = 86%	Internal
				KL2 = 69%	KL2 = 84%
				KL3 = 86%	KL3 = 98%
				KL4 = 85%	KL4 = 99%
Nguyen et al. (50)		0.79		53%		External (MOST)
Tiulpin & Saarakkala (51)		0.82	0.67			External (MOST)
Kim et al. (52)	KL0, 1 = 0.80					Internal
	KL2 = 0.69
	KL3 = 0.89
	KL4 = 0.95
Thomas et al. (53)		0.86	0.66			Internal (OAI)
Brejnebøl et al. (54)	0.84	0.88	KL0 = 79%	KL0 = 100%	KL0 = 76%	External
			KL1 = 73%	KL1 = 27%	KL1 = 98%
			KL2 = 83%	KL2 = 75%	KL2 = 85%
			KL3 = 92%	KL3 = 94%	KL3 = 91%
			KL4 = 96%	KL4 = 86%	KL4 = 100%

AUC = area under curve, KL = Kellgren-Lawrence, MOST = Multicenter Osteoarthritis Study, OAI = Osteoarthritis Initiative