Table 2

Best regression model obtained for the use of longitudinal data and specified input and output. The table shows Support Vector Machines SVM as the best model.

Longitudinal data
Given numberInputBest regression (from 19 trained)Output (at age+n)
3Age, gender, sphereSVR—Linear
R2: 0.57
RMSE: 0.09 D
AxL/Cr
4Age, gender, AxL/CrSVM—Quadratic
R2: 0.57
RMSE: 1.33 D
Sphere
5Age, gender, sphereSVR—Linear
R2: 0.77
RMSE: 0.97 D
Sphere
  • Axl/Cr, axial length to corneal radius ratio; D, diopter; R2, coefficient of determination; RMSE, root mean squared error; SVM-Quadratic, quadratic support vector machine; SVR, linear support vector regression.