Reporting of race and ethnicity in studies of artificial intelligence in pediatric urology: A secondary analysis of the AI-PEDURO online repository
Year: 2026
Type: review
Abstract: INTRODUCTION: While an increasing number of artificial intelligence (AI) models are being developed in pediatric urology, the extent of race/ethnicity reporting among these studies is unclear. Our objective was to evaluate the inclusion and quality of race/ethnicity reporting in AI models in pediatric urology. METHODS: We conducted a secondary analysis of studies included in the AI in PEDiatric UROlogy (AI-PEDURO) collaborative living scoping review and repository. We examined racial/ethnic groups reported, their proportional representation, use of race/ethnicity as a predictor, conducting of stratified analyses by race/ethnicity, data collection methods, bias evaluation, and discussions of implications on equity. RESULTS: Of 81 studies in the AI-PEDURO repository, six (7.4%) reported race/ethnicity. Five studies included White and Black patients, representing 4824/7968 (60.5%) and 1377/7968 (17.3%) of the pooled cohort, respectively. Asian patients were included in three studies and represented 178/6861 (2.6%). Two studies reported Native Hawaiian or other Pacific Islander and Hispanic or Latino patients, representing 20/6704 (0.3%) and 1236/6704 (18.4%), respectively. One study included American Indian or Alaska Native patients, representing 69/6604 (1.0%). Mixed patients were included in three studies and represented 103/7711 (1.3%). Race/ethnicity was a predictor variable in 4/6 studies. None of these six studies conducted stratified analyses of model performance across race/ethnicity subgroups, reported race/ethnicity data collection methodologies, examined algorithmic biases, discussed implications on equity, or examined socioeconomic status or geographic residence. CONCLUSIONS: Race/ethnicity reporting is poor in most AI studies in pediatric urology. Standardized reporting may help ensure fairness and generalizability of models across diverse pediatric urology populations.
DOI: 10.5489/cuaj.9541
Authors: Hyunwoong Harry Chae, David-Dan Nguyen, Adree Khondker, Mandy Rickard, Lauren Erdman, Andrew T. Gabrielson, Jin Kyu Kim, Jethro CC. Kwong, Brian Chun, Tariq Abbas, Nicolas Fernandez, Katherine Fischer, Lisette A. 't Hoen, Daniel T. Keefe, Hsin-Hsiao (Scott) Wang, John Weaver, Armando J. Lorenzo, Caleb P. Nelson
Institutions: University of British Columbia, University of Toronto, Hospital for Sick Children, Cincinnati Children's Hospital Medical Center, University of Cincinnati, Johns Hopkins University, Johns Hopkins Medicine, Riley Hospital for Children, Children's Hospital Los Angeles, Sidra Medical and Research Center, Seattle Children's Hospital, Children's Hospital of Philadelphia, Erasmus MC, Erasmus MC - Sophia Children’s Hospital, Izaak Walton Killam Health Centre, Boston Children's Hospital, Cleveland Clinic
Language: English
Subfield: Health Informatics
Field: Medicine
Domain: Health Sciences
Open Access status: diamond
OpenAlex ID: W7143643627