Probabilistic Prediction for the Detection of Vesicoureteral Reflux

Vesicoureteral Reflux (VUR) is a pediatric disorder in which urine flows backwards from the bladder into one or both ureters and, in some cases, into one or both kidneys. This has potentially very serious consequences as in the case of a Urinary Tract Infection, which is the main symptom of VUR, bacteria have direct access to the kidneys and can cause a kidney infection (pyelonephritis). The principal medical examination for the detection of VUR is the voiding cysteourethrogram (VCUG), which is not only a painful procedure, but also demands the exposure of the child to radiation. In an effort to avoid the unnecessary exposure of children to radiation, this study examines the use of a novel machine learning framework, called Venn Prediction, for reliably assessing the risk of a child having VUR. Venn prediction is used for obtaining lower and upper bounds for the conditional probability of a given child having VUR. The important property of these bounds is that they are guaranteed (up to statistical fluctuations) to contain well-calibrated probabilities with the only requirement that observations are independent and identically distributed (i.i.d.).