Table 3

Model error for the prediction of the 1–7-year GH growth response expressed in Dheight SD scores for the model group and the validation group separately: for the total group and subdivided into children born AGA or SGA.
Model group
 Validation group
Total group:
 n
 SDres
 SDstudRes
 Total group:
 n
 SDres
 SDstudRes
     1–2 year
 162
 0.17
 1.00
      1–2 year
 205
 0.19
 1.11
     1–3 years
 97
 0.29
 1.00
      1–3 years
 111
 0.28
 0.97
     1–4 years
 53
 0.38
 1.00
      1–4 years
 58
 0.35
 0.90
     1–5 years
 34
 0.43
 1.00
      1–5 years
 40
 0.45
 1.04
     1–6 years
 17
 0.48
 1.00
      1–6 years
 29
 0.40
 0.84
     1–7 years
 7
 0.57
 1.00
      1–7 years
 9
 0.43
 0.75
AGA:


AGA:


     1–2 year
 125
 0.17
 0.99
      1–2 year
 117
 0.19
 1.12
     1–3 years
 69
 0.28
 0.96
      1–3 years
 63
 0.26
 0.90
     1–4 years
 39
 0.39
 1.02
      1–4 years
 31
 0.31
 0.82
     1–5 years
 21
 0.44
 1.03
      1–5 years
 18
 0.39
 0.92
     1–6 years
 8
 0.45
 0.95
      1–6 years
 15
 0.42
 0.88
     1–7 years
 7
 0.57
 1.00
      1–7 years
 3
 0.26
 0.45
SGA:


SGA:


     1–2 year
 37
 0.18
 1.05
      1–2 year
 88
 0.19
 1.09
     1–3 years
 28
 0.31
 1.08
      1–3 years
 48
 0.30
 1.05
     1–4 years
 14
 0.36
 0.93
      1–4 years
 27
 0.37
 0.98
     1–5 years
 13
 0.41
 0.96
      1–5 years
 22
 0.49
 1.13
     1–6 years
 9
 0.50
 1.04
      1–6 years
 14
 0.37
 0.79
     1–7 years
 0
 -
 -
      1–7 years
 6
 0.49
 0.87

SDres = root mean square error of the residuals.

SD stud Res = SDres/SDresModelgroup

For a reliable model, the SDres should be similar, or lower, for the validation group compared with the SDres from the model group.

Kriström et al. BMC Medical Informatics and Decision Making 2009 9:1   doi:10.1186/1472-6947-9-1