Real-life benefit of artificial intelligence-based fracture detection in a pediatric emergency department

RBfracture v1.8 achieved a sensitivity of 92%, specificity of 83%, and overall accuracy of 87%. When supported by RBfracture, radiology residents improved across all metrics—sensitivity increased from 84% to 87%, specificity from 91% to 92%, and diagnostic accuracy from 88% to 90%. The algorithm performed consistently well in children under and over the age of five, with no significant drop in performance.

Accuracy of radiologists and radiology residents in detection of paediatric appendicular fractures with and without artificial intelligence

RBfracture v2.0 demonstrated the highest standalone performance, with 97% accuracy and 97% sensitivity. Both junior and senior readers improved their diagnostic accuracy when supported by RBfracture. Notably, the AI successfully detected six subtle fractures that were missed by all human readers, highlighting its value in enhancing clinical decision-making.