Figure 1 | Scientific Reports

Figure 1

From: Detection of subclinical keratoconus using a novel combined tomographic and biomechanical model based on an automated decision tree

Figure 1

The CHAID model to discriminate between NE and SKC in the training database. This discriminating rule achieved 82.8% sensitivity and 92.9% specificity. When generating a split, the following cutoff values were obtained: BAD-D: 0.92, 1.53, and 1.94; DA-ratio, 4.5; SPA1, 92.2; and Becc, 0.6. Details of the splitting among various categories of eyes are specified in the boxes at each node. The category with the gray background was the targeted category. CHAID, Chi-square automatic interaction detection; NE, normal eye; SKC, subclinical keratoconus; BAD-D, Belin/Ambrósio deviation index; DA ratio, deformation amplitude ratio; SPA1, stiffness parameter at first applanation; Becc, back eccentricity.

Back to article page