Skip to main content
. 2023 Nov 9;13(1):237–249. doi: 10.1007/s40123-023-00844-4

Table 2.

Prediction performance for each formula in all patients (N = 328)

Formula PE ± SD (µm) MAE (µm) MedAE (µm) % of eyes within PE range a
 ± 50 µm (%)  ± 100 µm (%)  ± 200 µm (%)  ± 300 µm (%)
KS 46.38 ± 154.48 126.55 101.00 23.4 49.7 79.9 94.5
WH -41.65 ± 157.01 131.45 116.65 22.8 43.2 80.5 93.5
Luo -68.37 ± 157.97 142.01 123.62 20.4 42.3 75.7 91.7
NK 54.01 ± 185.08 156.20 141.50 19.8 36.7 68.9 88.2
Hun 174.00 ± 169.99 192.01 152.68 19.2 33.4 58.9 76.6
ZZ 172.01 ± 234.87 237.48 196.00 13.0 23.4 50.9 70.3
Zhu 215.46 ± 191.47 241.23 225.98 10.7 23.7 45.3 65.7

The PE, MAE, MedAE, SD, and percentage of eyes with PE within ± 50 µm, ± 100 µm, ± 200 µm, and ± 300 µm for each of the seven formulae. The best MedAE value was found for KS (101.00 µm) and WH (116.65 µm) formulae; the worst result was produced by the ZZ (196.00 µm) and Zhu (225.98 µm) formulae

a = Cochran’s Q test was run to determine whether there were differences in proportion of eyes with a PE within ± 50 µm, ± 100 µm, ± 200 µm, and ± 300 µm of the formulae included. PE within ± 50 µm, ± 100 µm, ± 200 µm, and ± 300 µm were statistically significantly different, P < 0.001

PE prediction error; SD standard deviation; MAE mean absolute prediction error; MedAE median absolute prediction error