Skip to main content

Table 2 The absolute prediction errors of different formulas using total keratometry or standard keratometry in highly myopic eyes

From: Application of total keratometry in ten intraocular lens power calculation formulas in highly myopic eyes

IOL formula SD of PE TK method (n = 103) P1 value for multiple comparisons SD of PE K method (n = 103) P2 value for multiple comparisons
  MAE ± Std MedAE MAE ± Std MedAE
XGB 0.518 0.379 ± 0.351 0.324 0.572 0.409 ± 0.398 0.296
RBF 3.0 0.527 0.411 ± 0.341 0.340 0.536 0.419 ± 0.352 0.355
Kane 0.536 0.518 ± 0.365 0.435 P1 (vs. XGB) = 0.018* 0.539 0.527 ± 0.381 0.475
BUII 0.604 0.516 ± 0.401 0.435 P1 (vs. XGB) = 0.022* 0.630 0.536 ± 0.414 0.445 P2 (vs. XGB) = 0.021*
EVO 2.0 0.593 0.530 ± 0.400 0.440 P1 (vs. XGB) < 0.001*
P1 (vs. RBF) = 0.002*
0.591 0.532 ± 0.412 0.425 P2 (vs. XGB) = 0.002*
P2 (vs. RBF) = 0.005*
K6 0.737 0.579 ± 0.489 0.435 P1 (vs. XGB) = 0.017* 0.731 0.572 ± 0.487 0.415
HaigisWK 0.692 0.569 ± 0.465 0.460 P1 (vs. XGB) < 0.001*
P1 (vs. RBF) = 0.003*
0.686 0.563 ± 0.471 0.465 P2 (vs. XGB) < 0.001*
P2 (vs. RBF) = 0.002*
Haigis 0.712 0.589 ± 0.488 0.450 P1(vs. XGB) = 0.004*
P1 (vs. RBF) = 0.020*
0.715 0.584 ± 0.483 0.475 P2(vs. XGB) = 0.016*
P2(vs. RBF) = 0.040*
SRK/TWK 0.621 0.535 ± 0.401 0.495 P1 (vs. XGB) = 0.004*
P1 (vs. RBF) = 0.017*
0.629 0.544 ± 0.409 0.465 P2(vs. XGB) = 0.003*
P2(vs. RBF) = 0.009*
SRK/T 0.733 0.565 ± 0.470 0.460 P1 (vs. XGB) = 0.067* 0.720 0.550 ± 0.466 0.445
  1. TK = total keratometry; K = standard keratometry; SD = standard deviation; PE = prediction error; MAE = mean absolute error; Std = standard deviation; MedAE = median absolute error; XGB = XGBoost enhancement calculator; RBF 3.0 = Radial Basis Function formula version 3.0; BUII = Barrett Universal II formula; EVO 2.0 = Emmetropia Verifying Optical formula version 2.0; K6 = Cooke K6 formula
  2. P1: P value of the Bonferroni correction for multiple comparisons with all TK method
  3. P2: P value of the Bonferroni correction for multiple comparisons with all K method
  4. *Statistically different (P < 0.05)