Research Article

An Artificial-Intelligence- and Telemedicine-Based Screening Tool to Identify Glaucoma Suspects from Color Fundus Imaging

Table 2

Accuracy, sensitivity, specificity, Cohen’s kappa, and AUC for the system with 95% confidence intervals on the test data and ORIGA-light (validation data), with the ratio over 0.5 considered as a positive case for the model.

MeasureTest data (with 95% CI)ORIGA-light (with 95% CI)

Accuracy89.67% (85.65% to 92.87%)80.11% (73.64% to 85.59%)
Sensitivity83.33% (75.44% to 89.51%)84.96% (81.32% to 88.12%)
Specificity93.89% (89.33% to 96.91%)83.54% (80.43% to 86.34%)
Cohen’s kappa0.78 (0.71 to 0.85)0.62 (0.57 to 0.67)
Area under the curve0.93 (0.89 to 0.96)0.85 (0.81 to 0.90)