Abstract
Background
Foreign body injuries in children represent a critical public health challenge, with both food and non-food objects posing serious risks. Traditional risk assessment methods often overlook detailed morphological characteristics that can inform the risk of airway or gastrointestinal lodging. This study harnesses the potential of advanced AI and 3D structured-light scanning technologies to derive comprehensive risk profiles of objects implicated in pediatric injuries.
Methods
Data were sourced from the Susy Safe registry, which documents over 37,000 cases of foreign body incidents. Each object undergoes a standardized 3D scanning protocol, including pre-scan documentation, photographic imaging, and high-resolution multi-angular scanning. Extracted morphometric parameters—such as maximum diameter, length-to-width ratio, sphericity, surface roughness, and volume—are analyzed using computer-aided design (CAD) tools. Machine learning algorithms then integrate these data with detailed clinical and demographic information to develop predictive risk models. The resulting insights are integrated within a web-based platform that facilitates real-time risk assessment and supports clinical decision-making.
Results
The AI-driven pipeline successfully identified key geometric and material features of hazardous objects, establishing dimensional risk thresholds in relation to pediatric anatomical benchmarks. The predictive models demonstrated promising accuracy in stratifying the risk for airway and gastrointestinal lodging based on object profiles. This evidence-based approach enables targeted prevention strategies by flagging high-risk items and providing actionable insights to caregivers and health professionals.
Conclusions
Integrating advanced AI methods with cutting-edge 3D imaging technology delivers a robust framework for injury risk profiling in children. The developed platform not only facilitates real-time clinical assessment but also enhances public awareness.
Key messages
• Advanced AI and 3D imaging can identify hazardous objects through detailed risk profiling.
• Integrating morphometric and clinical data improves predictive accuracy for pediatric injuries.
Topic
3D Imaging, Artificial Intelligence, Pediatric Safety.