Hydronephrosis Severity Index

This study developed a deep learning–based Hydronephrosis Severity Index (HSI) that analyzes pediatric renal ultrasound images to predict the need for surgical intervention. The model showed high accuracy across multiple hospitals, achieving over 90% AUROC and strong sensitivity and specificity in both internal and external validation sets. Overall, the HSI can reliably assess hydronephrosis severity from a single ultrasound and help clinicians identify high-risk patients while safely reducing unnecessary follow-up for low-risk cases.