Machine-learning prediction model for personalized urinary tract infection care in children

Key Words: Machine-learning, artificial intelligence, targeted therapy for children, personalized treatment, high-risk urinary tract infection

Unlike in adults, urinary tract infection (UTI) in children is a silent and dangerous condition due to high risk of progression to severe infection and permanent loss of kidney function. Special challenges can make it hard to diagnose in young children, including the difficulty in obtaining urine samples. Timely identification of abnormal anatomy is essential to the long-term health for children.

Our team has identified multiple care gaps in managing UTI in children across multiple health systems: many children do not receive any testing even after the initial serious UTI, and the timing for those who receive care is often inconsistent. This can lead to conflicting management of this common condition, with delays in diagnosis, unnecessary testing, or even life-threatening complications, including severe infections and permanent kidney injury.

This project aims to mend this gap with cutting-edge innovation in artificial intelligence methodology. We seek to keep clinicians and parents informed of the dangers of UTI, as well as provide individualized risk prediction to encourage shared provider-patient decision-making for consistent care. By performing this research at Boston Children’s Hospital and the extensive network of partnering pediatric clinics, we will be able to trial this new approach directly with a large patient population. We will learn how to best deliver the prediction results to clinicians and families in need, as our ultimate goal is to improve outcomes for children with UTI. With the conceptual and technological innovation from this project, we expect the result of this study will change the status quo by proving a clear roadmap that allows us to translate cutting-edge research to routine clinical practice. This work has broad impacts and can be scaled to other institutions and conditions, facilitating interactive improvements that empower both clinicians and caregivers to meet the diverse clinical needs among children.