대한신장학회

논문분류	춘계학술대회 초록집
제목	Machine Learning Based Early Risk Stratification in Pediatric Nephrotic Syndrome: A Comprehensive Review
저자	Praveen Ram Elango
출판정보	2026; 2026(1):
키워드	Machine learning, pediatric neprhotic syndrome, Random forest
초록	Objectives: To systematically evaluate machine learning (ML) algorithms used for risk prediction and outcome stratification in pediatric idiopathic nephrotic syndrome. To analyze the data architectures and predictive feature sets including clinical, laboratory, and ancillary variables models and assess their impact on model performance. To identify translational challenges, including generalizability, interpretability, and clinical integration barriers, that limit routine implementation of ML-based decision support in pediatric nephrotic syndrome management Methods: A systematic search was performed from multiple databases including PubMed, Reference Citation Analysis, Science Direct, and Google Scholar. The search terms “Pediatric nephrotic syndrome” and “Machine learning” were used to identify relevant studies published in pediatric population. Data were critically analyzed and summarized descriptively. Results: Five machine learning algorithms, including extreme gradient boosting (XGBoost), gradient boosting decision tree (GBDT), extremely random trees (ET), random forest (RF), and logistic regression (LR) were used for model generation and validation. High predictive performance was demonstrated by the studies, with the random forest model outperforming the others with an accuracy of 99.14%, precision of 99.13%, recall of 99.16%, F1 score of 0.9914, and AUC value of 0.9983 on the validation set. The reliability and clinical interpretability of the model were confirmed by the SHAP analysis results, which demonstrated that markers including plasma IgG, total protein, complement C3, and ASO titer greatly contributed to model prediction and were highly congruent with the clinical pathological mechanism of NS. Additionally, RF and ET models showed good AUC values (0.94–0.95) and accuracy (80%) in predicting treatment responses, such as frequent relapses or steroid resistance. Conclusion: High prediction accuracy and significant clinical interpretability are demonstrated by machine learning-based models, offering compelling evidence for early risk assessment and customized treatment plans in nephrotic syndrome. The Random Forest (RF) model outperformed the other assessed methods.
원문(PDF)	PDF 원문보기

간행물 검색

대한신장학회 회원관리 규정