AIPEDURO

Kabir et al. (2023)

May 15, 2024

Written by:

VUR grade, Vesicoureteral Reflux and Urinary Tract Infection

Article DOI	https://doi.org/10.1016/j.jpurol.2023.10.030
Objective	To predict VUR grade from VCUG
AI Approach	Multiple classification algorithms (MLP, Extra Trees, Random Forrest, Gradient Boosted Trees, SVM)
Data Source(s)	Online image scraping (113 VCUG images)
Model Input	VCUG Image, Nine input features
Model Outcome	Binary VUR grade (low [I-III] vs. high [IV-V])
Model Metrics	F1 = 90.3% (SVM) F1 = 91.1 % (MLP) AUROC = 0.96
Model Usability	NA

AI = Artificial Intelligence, SVM = Support Vector Machines, MLP = Multi-Layer Perceptron, AUROC = Area under receiver operator characteristic