Diagnostic performance of artificial intelligence in pediatric diseases: a systematic review and meta-analysis

Background Artificial intelligence (AI) has increasingly been applied in medical diagnostics and has shown promise in improving the detection of pediatric diseases. This systematic review and meta-analysis evaluated the diagnostic performance of AI models in identifying pediatric respiratory, systemic/inflammatory, and neurological conditions, specifically pneumonia, tuberculosis, Kawasaki disease, and brain tumors. Methods This study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis for Diagnostic Test Accuracy Studies (PRISMA-DTA) guidelines. PubMed, EMBASE, Web of Science, and IEEE Xplore were searched from inception to March 2026. Studies were included if they evaluated the diagnostic performance of AI models for pediatric diseases. Methodological quality was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool. Diagnostic performance indicators—sensitivity and specificity—were pooled using a bivariate random-effects model. Results AI models demonstrated high diagnostic performance for pneumonia, with pooled sensitivity and specificity of 93% and 90%, respectively. AI also showed excellent performance in detecting Kawasaki disease, with sensitivity and specificity of 94% and 89%, respectively. Studies assessing brain tumor detection reported strong performance; however, inconsistent reporting of metrics prevented quantitative meta-analysis. In contrast, AI models showed low sensitivity (41%.) in detecting tuberculosis. Conclusion AI models show strong potential to improve diagnostic accuracy in pediatric diseases, particularly pneumonia and Kawasaki disease. However, the lack of external validity, high risk of bias, and substantial between-study heterogeneity limits generalizability. Therefore, future research should prioritize external validation across diverse datasets to ensure reliability and support integration of AI-based diagnostic tools into pediatric clinical practice.