Logistic Regression Model Using Scheimpflug-Placido Cornea Topographer Parameters to Diagnose Keratoconus

Table 5

Performance of the logistic regression analysis models for differentiating keratoconus eyes from normal eyes.

Model	Sensitivity	Specificity	−2 log-likelihood
Model 1	%91.2	%95.8	94, 624
Model 2	%89, 6	%97, 5	98, 971
Model 3	%91, 2	%98, 3	84, 663
Model 4	%91, 2	%97, 5	79, 305
Model 5	%93, 6	% 98, 3	59, 774
Model 6	%94, 4	% 98, 3	44, 416
Model 7	%96, 8	%99.2	47, 993
Model 8	%96, 8	%99.2	42, 461

Model 1 = anterior Ø = 3 mm K2 + posterior Ø = 3 mm K2 + K_max + CCT + TCT; Model 2 = anterior rf + anterior rs + anterior Q + anterior RMS/A; Model 3 = posterior rf + posterior rs + posterior Q + posterior RMS/A; Model 4 = Model 2 + Model 3; Model 5 = BCV_f + BCV_b + KV_f + KV_b + SI_f + SI_b; Model 6 = all of the aberrations on Table 2; Model 7 = BCV_f + BCV_b + posterior rf + KVf + anterior coma Z (3, ±1); Model 8 = BCV_f + BCV_b + posterior rf + TCT + CCT.