Table 2 Accuracy and AUC of fracture diagnosis and fracture classification in included studies

Adams [11]

88.1–94.4 (AI)

93.5 (specialist)

92.9 (residents)

90.5 (AI + medically naïve)

87.6 (medically naïve)

0.94–0.98 (AI)

Urakawa [12]

95.5 (AI)

92.2 (human)

0.984 (AI)

0.969 (human)

Cheng [13]

91 (AI)

0.98 (AI)

Krogue [14]

93.7 (AI)

0.975 (AI)

91.2 (AI)

0.873–1.00 (AI)

Yu [15]

96.9 (AI)

0.9944 (AI)

93.9–98.5 (AI)

0.95–0.99 (AI)

Mutasa [16]

92.3 (AI)

0.92 (AI)

86 (AI)

0.96 (AI)

Beyaz [17]

79.3 (AI)

Mawatari [18]

0.832 (human)

0.905 (AI)

0.876 (AI + human)

Yamada [19]

98 (AI)

Cheng [10]

92.67(AI)

97.1 (AI + human)

Yoon [8]

97 (AI)

90 (AI)

Sato [1]

96.1 (AI)

84.7 (human)

91.2 (AI + human)

0.99(AI)

Bae [21]

97.1 (AI)

0.977 (AI)

Murphy [7]

77.5 (human)

92 (AI)

0.98 (AI) for normal

0.99 (AI) for neck Fx

0.97(AI) for ITC Fx