Table 2. Emergency doctor and YOLOv8 model performance comparison based on the test datasets.

Emergency doctor
	Without artificial intelligence
		Total sets	True positive	True negative	False positive	False negative	Sensitivity (95%CI)	Specificity (95%CI)	Accuracy (95%CI)
	All	1,000	655	230	71	44	93.7%	75.2%	88.0%
	Fracture	400	225	118	38	19	92.2%	75.6%	85.8%
	Not fracture	600	430	112	33	25	94.5%	77.2%	90.3%
	With artificial intelligence
		Total sets	True positive	True negative	False positive	False negative	Sensitivity (95%CI)	Specificity (95%CI)	Accuracy (95%CI)
	All	1,000	688	261	30	21	97.0%	89.7%	94.9%
	Fracture	400	239	132	21	8	96.8%	86.3%	92.8%
	Not fracture	600	449	129	9	13	97.2%	93.5%	96.3%