Prediction accuracy of adult stature through the use of Japanese athletes' height data in the BTT Model

Abstract

The purposes of this study were to examine the Japanese parameters (JP) of the BTT model (Bock, 1992) and measure the accuracy of the predicted adult stature (PAS) estimation using the height data of Japanese children (6 to 17 years old). The subjects of this study were 1,508 boys and 1,308 girls. Among them, 1,137 boys and 1,102 girls with good fitting on the BTT model were analyzed. BTT parameters were created for Japanese subjects and compared to default parameters (DP), followed by an error estimation comparison between PAS using JP and PAS using DP.Additionally, growth data was artificially generated by subtracting the newest year from the 12-year data and calculating PASs from that data set. Finally, we compared the difference between the PAS of the growth data and the PAS from the full 12 years of data. The following findings were obtained from the analyzes: 1) Initial parameters for Japanese (JP) in the BTT model of height development for predicting adult heights of athletes were obtained for males and females separately.2) Highly accurate predictions can be made when using 8 years of developmental data from 6 to 13 years of age, but the prediction accuracy is significantly reduced when the developmental data is less than 13 years of age.3) In stunting data, there is a regression effect whereby a lower PAS is predicted for taller individuals and conversely a higher PAS is predicted for shorter individuals. 4) In the growth data, both boys and girls with developmental characteristics such as abrupt stalls in growth and growth spurts resulted in an estimation error of ± 10 cm or more.