Table 7 Pairwise comparison of the prediction accuracy (i.e., the percentage of eyes within a given range of prediction error) between the MM formula and each of the other four formulae (SRK/T, Hoffer Q, Holladay I and Haigis) for the different types of eyes (long, long medium, medium, short and all eyes), using the McNemar test at a statistical significance level of 5%.

From: An ensemble-based approach for estimating personalized intraocular lens power

Monofocal Lenstec Softec HDO

Pairwise comparison of the prediction accuracy

p-values for McNemar test

Long eyes

 

% Rx within ± 0.5D

% Rx within ± 1.0D

% Rx within ± 1.5D

MM vs SRK/T

 < 0.001b

 < 0.001b

 < 0.001b

MM vs Hoffer Q

 < 0.001b

0.999a

0.999a

MM vs Holladay I

 < 0.001b

0.999a

0.999a

MM vs Haigis

0.041b

0.999a

0.999a

Long medium eyes

 

% Rx within ± 0.5D

% Rx within ± 1.0D

% Rx within ± 1.5D

MM vs SRK/T

 < 0.001b

 < 0.001b

 < 0.001b

MM vs Hoffer Q

 < 0.001b

0.789a

0.999a

MM vs Holladay I

0.211a

0.999a

0.999a

MM vs Haigis

0.864a

0.752a

0.999a

Medium eyes

 

% Rx within ± 0.5D

% Rx within ± 1.0D

% Rx within ± 1.5D

MM vs SRK/T

 < 0.001b

 < 0.001b

0.999a

MM vs Hoffer Q

 < 0.001b

 < 0.001b

0.999a

MM vs Holladay I

 < 0.001b

 < 0.001b

0.617a

MM vs Haigis

 < 0.001b

 < 0.001b

0.999a

Short eyes

 

% Rx within ± 0.5D

% Rx within ± 1.0D

% Rx within ± 1.5D

MM vs SRK/T

0.003b

 < 0.001b

 < 0.001b

MM vs Hoffer Q

0.585a

0.770a

0.041c

MM vs Holladay I

0.001b

0.607a

0.041c

MM vs Haigis

0.831a

 < 0.001c

0.041c

All eyes

 

% Rx within ± 0.5D

% Rx within ± 1.0D

% Rx within ± 1.5D

MM vs SRK/T

 < 0.001b

 < 0.001b

 < 0.001b

MM vs Hoffer Q

 < 0.001b

0.013b

0.023c

MM vs Holladay I

 < 0.001b

0.013b

0.343a

MM vs Haigis

 < 0.001b

0.391a

0.023c

  1. aNo statistically significant difference at level 0.05.
  2. bMM formula outperformed at significance level 0.05.
  3. cMM formula underperformed at significance level 0.05.
  4. Significant values are in [bold, Italics].