Artificial Intelligence in Pediatric Bronchoscopy: Current Evidence and Future Perspectives

BACKGROUND: Artificial intelligence (AI) is rapidly advancing in medical endoscopy, with proven value in adult bronchoscopy for diagnostic enhancement and training. Pediatric bronchoscopy, with its unique anatomical and procedural challenges, is beginning to explore AI integration. This narrative review summarizes current evidence and anticipates future applications. METHODS: We searched PubMed and EMBASE for articles in English from inception to August 2025 using combinations of the terms "pediatric bronchoscopy," "artificial intelligence," "machine learning," and "deep learning." Additional relevant publications were identified by reviewing reference lists and conference abstracts. Inclusion was based on relevance to AI applications in pediatric or, when pediatric data were lacking, adult bronchoscopy. Areas of emphasis included AI-driven video image analysis, navigation guidance, diagnostic decision support, and training tools, along with challenges in data annotation, bias, and generalizability. RESULTS: Deep learning models can identify airway anatomy on bronchoscopy video with expert-level accuracy and, when integrated into simulators, improve novices' completeness, procedural structure, and speed. AI applied to chest radiographs shows high accuracy in predicting foreign body aspiration, potentially reducing unnecessary bronchoscopies. A validated pediatric bronchitis scoring tool standardizes airway inflammation assessment, supporting future AI automation. Limitations include small, single-center datasets and the need for pediatric-specific data to prevent bias. CONCLUSIONS: AI in pediatric bronchoscopy is at an early stage, with most evidence extrapolated from adult practice. Available data suggest potential to improve diagnostic accuracy, safety, and training efficiency. Progress will require multi-center pediatric data collaboration, prospective trials, and workflow integration.