Table 4 Performance metrics for different models in identifying eyes at risk of surgery for uncontrolled glaucoma.

From: Deep learning-based identification of eyes at risk for glaucoma surgery

Time horizon (years)

Logistic regression

AUC (95% CI)

Neural network

AUC (95% CI)

DLM

AUC (95% CI)

0–0.25

0.83 (0.77, 0.88)*

0.86 (0.81, 0.91)*

0.92 (0.88, 0.96)

0.25–0.5

0.83 (0.73, 0.93)*

0.86 (0.73, 0.93)*

0.92 (0.85, 0.99)

0.5–1

0.81 (0.72, 0.89)*

0.85 (0.77, 0.92)*

0.88 (0.77, 0.92)

1–2

0.74 (0.67, 0.82)*

0.79 (0.72, 0.86)*

0.84 (0.78, 0.90)

2–3

0.70 (0.62, 0.79)*

0.75 (0.67, 0.83)*

0.83 (0.76, 0.90)

3–4

0.68 (0.58, 0.79)*

0.73 (0.63, 0.83)

0.78 (0.68, 0.87)

4–5

0.68 (0.54, 0.82)*

0.72 (0.58, 0.85)

0.77 (0.63, 0.89)

  1. A comparison of AUC between models to determine if performance differences were statistically significant (p < 0.05) using the DeLong Test.
  2. *p \(<\) 0.05 when comparing the model AUC to the DLM at the same time horizon.
Back to article page