Skip to main content
. 2022 Jun 22;14:891202. doi: 10.3389/fnagi.2022.891202

TABLE 4.

Median absolute error and its standard error of the RFR and GLM model for the monosyllabic score at a presentation level of 65 dBSPL (WRS65), the maximum word recognition score (WRSmax) and the presentation level for the maximum word recognition score, Lmax.

Target Cost function Subgroup WHO0 WHO1 WHO2 WHO3 WHO4
WRS65 MAE
(percentage points)
Training RFR GLM 1.03 ± 0.04
1.91 ± 0.01
5.60 ± 0.06
9.04 ± 0.04
8.01 ± 0.11
14.36 ± 0.17
0.06 ± 0.02
2.09 ± 0.04
0.004 ± 0.002
0.018 ± 0.001
Test RFR GLM 1.53 ± 0.08
1.91 ± 0.01
8.88 ± 0.24
9.04 ± 0.24
12.70 ± 0.59
14.42 ± 0.79
0.10 ± 0.05
2.10 ± 0.11
0.004 ± 0.002
0.018 ± 0.001
WRSmax Training RFR GLM 0.04 ± 0.01
0.57 ± 0.01
1.54 ± 0.02
3.04 ± 0.03
5.65 ± 0.08
8.71 ± 0.08
10.79 ± 0.36
18.92 ± 0.41
6.76 ± 0.27
8.06 ± 019
Test RFR GLM 0.06 ± 0.01
0.57 ± 0.01
2.27 ± 0.05
3.04 ± 0.07
9.06 ± 0.25
8.71 ± 0.26
17.24 ± 1.94
18.91 ± 1.51
11.28 ± 1.56
7.82 ± 0.78
Lmax MAE
(dB)
Training RFR GLM 3.44 ± 0.04
5.31 ± 0.04
3.26 ± 0.05
5.30 ± 0.08
3.39 ± 0.06
5.85 ± 0.06
3.14 ± 0.13
5.81 ± 0.11
2.75 ± 0.16
8.63 ± 0.25
Test RFR GLM 5.09 ± 0.17
5.31 ± 0.18
5.26 ± 0.15
5.34 ± 0.23
5.41 ± 0.23
5.85 ± 0.30
5.16 ± 0.46
5.85 ± 0.55
4.42 ± 0.46
8.65 ± 1.15