Early stature prediction method using stature growth parameters

Abstract

Background: The creation of an accurate growth prediction method for human stature at a stage of growth has been an interesting challenge in medical science and human biology. Aim: The aim of this study was to develop a non-radiographic final stature prediction method that is applicable in the early pubertal growth period. Subjects and methods: Randomly selected 12-year serial stature growth data for 400 Koreans were fitted with two nonlinear growth curves: Preece and Baines model 1 (PB1) and Jolicoeur-Pontier-Pernin-Sempe (JPPS) functions. Five biological parameters, including take-off (TO) related parameters, were derived by differentiation of the two curves, respectively. Those five variables were composed into a multiple linear regression equation for final stature prediction. In the cross-validation subjects, TO-related variables were estimated by linear interpolation from the partial growth data prior to estimation age, then incorporated into the prediction equation. Results: The final stature prediction model had excellent validity and accuracy when applied to the cross-validation samples. Prediction accuracy increased according to increasing years after take-off. Conclusions: This study suggests that a final stature prediction method using multiple regression analysis that includes biological parameters can predict stature growth with sufficient validity and accuracy. Incorporation of TO-related parameters allowed us to develop earlier growth evaluation and prediction methods compared with other previous methods.