Answers are grounded in this paper and cite their sources.
Background Cyclosporine A (CsA) nephrotoxicity is a dose-limiting toxicity in pediatric immunosuppression, and early identification of children at risk of CsA-associated acute kidney injury (AKI) remains an unmet clinical need.Research design and methods In retrospective cohorts of CsA-treated inpatients undergoing therapeutic drug monitoring (352 children; 342 adults), a regularized logistic-regression model was compared with six machine-learning classifiers under five feature-selection schemes, three small-sample strategies were evaluated, and adult–child transferability was assessed on 95 aligned features. A 14-variable model was externally validated in an independent pediatric cohort (n = 257).Results Regularized logistic regression achieved the best discrimination in children (hold-out AUC 0.800) and adults (hold-out AUC 0.956), with complex models overfitting. Adult and pediatric risk architectures differed fundamentally, with transfer AUCs near chance (≈0.6). The 14-variable model retained discrimination on external validation (AUC 0.784, 95% CI 0.656–0.912).Conclusions On small clinical datasets, regularization rather than complexity determines predictive performance. We provide an interpretable, TDM-integrated pediatric CsA-AKI model, deployed as an online calculator, supporting early risk stratification in pediatric CsA therapy.
No takes yet. Share an insight, caveat, or question.
Zhou et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: