20Aug 2023
Obstruction risk from common ultrasound parameters in pediatric hydronephrosis with machine learning

Obstruction risk from common ultrasound parameters in pediatric hydronephrosis with machine learning

A new study has developed an AI model, called AERO (AI Evaluation of Renogram Obstruction), to predict the risk of renal obstruction in patients with isolated hydronephrosis using routine ultrasound findings. The study used data from 304 patients and found that AERO outperformed traditional SFU grade in predicting obstruction risk, achieving a binary AUROC of 0.84. This AI tool, available as an easy-to-use application, could potentially help 66 more patients per 1000 avoid unnecessary diuretic renography without missing significant obstruction if a 30% probability threshold is used. Further validation is needed before clinical implementation.

  • #urology

Like

Save

Share