Clinical translation of artificial intelligence in hypospadias: a systematic review and meta-analysis.

Artificial intelligence (AI) has emerged as a promising tool to improve the objectivity and reproducibility of hypospadias assessment. However, evidence regarding its clinical applications remains fragmented. This systematic review and meta-analysis evaluated the current role, diagnostic performance, and status of clinical translation of AI in hypospadias. A systematic search of PubMed/MEDLINE, Embase, and Scopus was performed according to PRISMA 2020 guidelines (PROSPERO: CRD420261296235). Studies evaluating AI applications in pediatric hypospadias were included. AI applications were categorized into penile curvature assessment, hypospadias classification, urethral plate assessment, and patient education. Random-effects meta-analyses were performed for clinically comparable studies. Sixteen studies published between 2021 and 2026 involving sample sizes ranging from 7 to 1,169 were included. Image-based computer vision and deep learning applications accounted for 62.5% of studies, whereas 37.5% evaluated large language models (LLMs). AI-assisted hypospadias classification demonstrated a pooled accuracy of 88% (95% CI, 87%-90%). AI-assisted penile curvature assessment achieved a pooled mean absolute error of 7.89° (95% CI, 6.43°-9.34°), while AI-generated educational tools demonstrated a pooled accuracy of 73% (95% CI, 60%-86%). Automated urethral plate assessment showed excellent performance, with landmark localization precision of 99.5% and sensitivity of 99.1% in the largest study. Risk of bias was generally low in the predictor and outcome domains but frequently high or unclear in the analysis domain because of limited external validation and methodological heterogeneity. Current evidence indicates that most AI applications in hypospadias remain at the proof-of-concept or early validation stage. Although computer vision has demonstrated promising performance for objective anatomical assessment and LLMs show potential for patient education, prospective clinical implementation remains limited and requires larger multicenter validation studies, standardized reporting, and prospective clinical evaluation before routine integration into clinical practice.