A prediction model could foresee adequate height response in children eligible for growth hormone treatment

Abstract

Many years of experience with recombinant growth hormone (GH) treatment has shown how effective it is in improving adult height in children with short stature due to GH deficiency (GHD). For those with severe GHD or severe GH insensitivity the diagnosis is often clear, but for a number of short patients it is difficult to draw a clear line between mild GHD, partial GHD and idiopathic short stature. 1 There is consensus that diagnosing GHD in childhood is a multifaceted process requiring both clinical and auxological data, as well as biochemical testing and radiological imaging of skeletal maturity and visualising the pituitary size or possible malformation. 2 However, there is still no gold standard diagnostic test for GHD. The provocation tests available are generally non-physiological and invasive, and are subject to variability between different assays. There is a debate whether measurement of spontaneous GH secretion is

Curve matching to predict growth in patients receiving growth hormone therapy: An interpretable & explainable method

“…A ‘one-size-fits-all’ approach (group-level models) thus does not work for all patients or healthcare systems. Furthermore, prediction models are known to underestimate relatively low predictions and overestimate relatively high predictions ( 18 , 19 ); the accuracy of a high prediction of height velocity is less than the accuracy of a low prediction ( 19 ), and this information is generally not presented to HCPs, once again hindering the implementation of prediction models in clinical practice. Finally, in order to be adopted by HCPs, the applied methods should be interpretable ( 20 ) and explainable in order for HCPs to understand, trust, and use the results.…”