Introduction: Growth retardation after renal transplantation in children is affected by both genetic factors and immunosuppressive regimens. In this study, we integrated genomics and clinical intervention strategies to construct a growth prediction model and explore the optimal effect of hormone withdrawal timing on postoperative growth pattern. 
Methods: A total of 729 height related single nucleotide polymorphisms (SNPS) were screened by whole exome sequencing. Combined with clinical variables such as age, seven machine learning algorithms including random forest were used to construct a growth pattern prediction model (△HAZ≥0.5 was defined as catch-up growth). The results were validated by 39 independent cohorts (2021-2022).

The effects of glucocorticoid withdrawal in 214 children (including the above cohort) were analyzed simultaneously. According to whether glucocorticoids were withdrawn within 3 months after surgery, the children were divided into glucocorticoid withdrawal group (72 cases) and non-glucocorticoid withdrawal group (142 cases), and the change rate of HAZ score was dynamically evaluated. 
Results: The genetic model showed that the random forest algorithm had the best prediction performance, with an AUC of 0.924 (accuracy 81.25%) in the training set and 0.796 (accuracy 79.49%) in the external validation set. Age and 19 SNPS (e.g., NUCB2 rs757081, PCSK1 rs6235) were the core predictors. Analysis of clinical interventions showed that: The HAZ score of the steroid withdrawal group was significantly improved at 12 months after surgery (-0.51±1.10 vs. -0.95±1.31, P=0.046), and the growth rate of the steroid withdrawal group was 98% higher than that of the non-steroid withdrawal group (0.076±0.135 vs. 0.038±0.078, P=0.016). It does not increase the risk of rejection. 
Conclusions: SNP-clinical features based machine learning models can accurately identify postoperative growth potential and guide individualized immune management (such as the timing of hormone withdrawal and growth hormone application). Early steroid withdrawal (within 3 months after surgery) can effectively prolong the growth acceleration period and improve catch-up efficiency. The combination of the two methods provides an integrated solution of "prediction-intervention" for growth disorders after renal transplantation in children.