AI Diagnostic Tools in Pediatrics: A Systematic Review of Accuracy, Safety, and Future Directions

Abstract

Background: Artificial intelligence (AI) is increasingly influencing healthcare by enhancing diagnostic accuracy and efficiency. In paediatrics, AI holds unique potential to address diagnostic delays, specialist shortages, and variability in care, particularly in under-resourced settings.

Objective: This systematic review explores the diagnostic accuracy, safety, and future directions of AI-based diagnostic tools in pediatric medicine.

Methods: A comprehensive literature search was conducted in PubMed, Embase, Scopus, and Web of Science. Studies have included AI applications in pediatric diagnostics across various settings, including clinical trials, retrospective analyses, systematic reviews, and case reports. Key outcomes evaluated were diagnostic performance, safety profiles, and implementation challenges.

Results: Seventy-four studies met the inclusion criteria. AI tools demonstrated high potential in pediatric diagnostics, especially in imaging, cardiology, infectious diseases, and developmental disorders. Diagnostic accuracy varied (AUC range: 0.78–0.96) based on algorithm type and dataset quality. Several studies reported enhanced diagnostic speed, reduced human error, and improved clinical decision-making. However, challenges such as algorithmic bias, lack of external validation, ethical concerns, and data standardisation were recurrent. Integration hurdles included interpretability, infrastructure limitations, and regulatory gaps.

Conclusion: AI-based diagnostic tools are reshaping pediatric healthcare by supporting clinicians and improving diagnostic precision. While promising, their safe and equitable implementation requires rigorous validation, ethical safeguards, and attention to health system integration. Future research should prioritise pediatric-specific models, real-world validation, and international consensus on AI governance.