Prediction of Final Stature from a New 3D Two-Stage Growth Model

Abstract

The purpose of the present study was to establish higher dimensional growth model and then to predict the final stature from a new 3D two-stage growth model. To check the validity of the models, a secondary longitudinal data on age, weight and stature of Japanese boys and girls were used. The proposed 3D growth model showed more precise than that of the existing 2D growth models. After extracting the final stature and stature at different ages from the well fitted proposed model and eliminating the problem of multicollinearity using ‘Forward Stepwise Ridge Regression’ and ‘Least Absolute Shrinkage and Selection Operator (LASSO)’ techniques, this study propose four different equations for predicting final stature with higher precession, validity and stability compared to others. These equations are perfectly applicable in Japanese population. But, the form of the proposed model and its procedures can be applied to others populations.