Abstract Background and Hypothesis Early identification and prediction of steroid-dependent/frequently relapsing nephrotic syndrome (SDNS/FRNS) in children have become increasingly important. The aim of this study is to develop and validate a model for the early identification and prediction of SDNS/FRNS in pediatric patients. Methods A multicenter retrospective study was conducted in China to investigate risk factors and develop a prediction model for the early identification of potential SDNS/FRNS in children with steroid-sensitive nephrotic syndrome (SSNS). Data were collected from 4 259 children diagnosed with SSNS at the Children’s Hospital of Chongqing Medical University between January 2009 and December 2023. After applying strict inclusion and exclusion criteria, 2 624 cases were included in this study. The cohort was divided into two groups: the control group, comprising 1 064 cases of Non SDNS/FRNS, and the experimental group, comprising 1 560 cases of SDNS/FRNS. Both groups were further split into training and validation sets at an 8:2 ratio. Additionally, data from other nationwide multicenter studies conducted between 2012 and 2021 were gathered for external validation. Results This study identified 15 acute-phase clinical predictors of SDNS/FRNS development. Among the eight tested models, the LightGBM showed optimal performance (AUC = 0.89) when these features were used. The simplified 15-variable LightGBM model maintained strong external validation accuracy (AUC = 0.93) and has been implemented in an online clinical tool for practical application. Conclusions This work presents the first machine learning-based early prediction model for SDNS/FRNS using clinical big data from the acute phase of the condition. Additionally, it introduces an online prediction tool for pediatric SDNS/FRNS.
Chen et al. (Mon,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: