Table 1 Sensitivity, specificity, and accuracy of the prototype CRADLE application (iPhone X) at detecting pathologic leukocoria in childhood photographs.

AgeSensitivity (TPR)Specificity (SPC)Accuracy (ACC)No. of
diagnosed
No./total no.*% (95% CI)No./total no.*% (95% CI)No./total no.*% (95% CI)
≤1 month3/1717.6 (0.0–35.8)10/1662.5 (38.8–86.2)13/3339.4 (22.7–56.1)3/20
≤2 months7/1741.2 (17.8–64.6)6/1637.5 (13.8–61.2)13/3339.4 (22.7–56.1)4/20
≤3 months10/1855.6 (32.6–78.5)7/1838.9 (16.4–61.4)17/3647.2 (30.9–63.5)6/20
≤6 months15/2075.0 (56.0–94.0)5/2025.0 (6.0–44.0)20/4050.0 (34.5–65.5)9/20
≤1 year18/2090.0 (76.9–100.0)4/2020.0 (2.5–37.5)22/4055.0 (39.6–70.4)14/20
≤1.5 years18/2090.0 (76.9–100.0)4/2020.0 (2.5–37.5)22/4055.0 (39.6–70.4)15/20
≤2 years18/2090.0 (76.9–100.0)4/2020.0 (2.5–37.5)22/4055.0 (39.6–70.4)20/20

*Total number refers to the total number of children whose faces and open eye(s) were photographed in each age interval. The natural occurrence of physiological leukocoria in control children lowers specificity over time. Sensitivity is low in the first few months of life because certain disorders, such as unilateral retinoblastoma, Coats’ disease, and refractive error, are typically not present at birth, whereas bilateral retinoblastoma can be present at birth or develop in the first few months of life. Number of children diagnosed in each respective age period.