Table 3 The diagnostic performance of residents with normal template- and AI-aided modes

From: Generative artificial intelligence for fundus fluorescein angiography interpretation and human expert evaluation

Conditions

Diagnostic performance

P value

Normal template-aided

AI-aided

Accuracy (%)

Sensitivity (%)

Specificity (%)

Accuracy (%)

Sensitivity (%)

Specificity (%)

(95% CI)

(95% CI)

(95% CI)

(95% CI)

(95% CI)

(95% CI)

AMD

92.07 (88.38 - 94.66)

80.85 (75.76 - 84.82)

94.24 (90.81 - 96.31)

93.45 (89.99 - 95.77)

89.36 (85.23 - 92.37)

94.24 (90.81 - 96.31)

<0.05

BRVO

96.55 (93.77 - 98.12)

81.82 (76.87 - 85.75)

98.44 (96.03 - 99.26)

97.59 (95.10 - 98.83)

90.91 (86.79 - 93.52)

98.44 (96.03 - 99.26)

<0.05

CRVO

95.52 (92.48 - 97.36)

70.83 (65.21 - 75.63)

97.74 (95.10 - 98.83)

97.24 (94.65 - 98.60)

83.33 (78.36 - 86.98)

98.50 (96.03 - 99.26)

<0.05

CSC

94.83 (91.64 - 96.84)

50.00 (44.28 - 55.72)

97.10 (94.21 - 98.36)

97.59 (95.10 - 98.83)

64.29 (58.47 - 69.44)

99.28 (97.00 - 99.65)

<0.05

DR

96.55 (93.77 - 98.12)

95.24 (92.06 - 97.10)

97.90 (95.10 - 98.83)

97.24 (94.65 - 98.60)

96.60 (93.77 - 98.12)

97.90 (95.10 - 98.83)

<0.05

VKH

99.31 (97.52 - 99.81)

71.43 (65.92 - 76.28)

100.00 (98.69 - 100.00)

97.93 (95.56 - 99.05)

14.29 (10.60 - 18.62)

100.00 (98.69 - 100.00)

<0.05

Normal

98.28 (96.03 - 99.26)

94.44 (90.81 - 96.31)

98.53 (96.03 - 99.26)

99.66 (98.07 - 99.94)

100.00 (98.69 - 100.00)

99.63 (97.52 - 99.81)

<0.05

Total

86.55 (82.14 - 90.00)

86.55 (82.14 - 90.00)

100.00 (98.69 - 100.00)

90.34 (86.40 - 93.24)

90.34 (86.40 - 93.24)

97.67 (95.10 - 98.83)

<0.05

  1. P value of per-disease diagnostic performance: McNemar’s Test.
  2. P value of overall diagnostic performance: Bowker’s Test of Symmetry.