[Mathematical models for predicting growth responses to growth hormone replacement therapy]

Growth prediction models are algorithms derived from multiple regression analyses including variables that influence growth responses to GH therapy in a defined group of subjects over a defined period of time. Mathematical equations can be derived from the knowledge acquired with the relative importance of each variable, which provide objective measurements of each subject's growth potential in response to GH therapy on different situations. Therefore, these equations can be used as tools to improve evidence-based decision regarding to growth promoting treatment strategies to be used in each child, optimizing cost-effectiveness with the lowest cumulative GH dose. Several models have already been developed to predict growth responses to GH for different short stature causes, but they still have low clinical usefulness, due to their low predictive power and low prevision accuracy. This has lead to a growing interest in the addition of new variables, such as biochemical or genetic markers, which could improve prevision accuracy and then allow, in the future, GH therapy individualization according to the specific needs of each child.