Reliability and validity of an artificial intelligence-assisted system for the detection of abnormalities in chest and bone radiographs in an emergency department
Emergencias, 2026
Abstract
Introduction. To evaluate the diagnostic performance for the detection of abnormalities of two commercial artificial intelligence (AI) systems—ChestView for chest radiographs (CXR) and BoneView for skeletal radiographs (SXR)—in a hospital emergency department (ED), and to compare their validity with that of observers with different professional profiles and levels of experience: emergency physicians, radiology residents, and expert radiologists.
Method. A diagnostic test evaluation study was conducted on a random sample of 346 CXRs and 261 SXRs requested in the emergency department. The examinations were independently analyzed by the AI systems and by the different observers. The reference diagnosis (gold standard) was established by consensus among three radiologists, using other available imaging studies or clinical information when necessary. Sensitivity, specificity, and positive and negative predictive values (NPV) were calculated and compared.
Results. For CXR, the AI system (ChestView) showed an overall sensitivity (64.4%) significantly higher than that of emergency physicians (49.2%; p = 0.018), although lower than that of the expert radiologist (83.9%; p < 0.001). Performance was notable for the detection of nodules/masses (sensitivity 80.0%) and pneumothorax (NPV 99.7%), but lower for consolidations (sensitivity 40.4%). For SXR, the AI system (BoneView) achieved a sensitivity for fracture detection (87.5%) higher than that of the expert radiologist (77.1%), with an NPV of 96.9%. However, its performance was lower for the detection of dislocations (sensitivity 60.0%) and joint effusions (sensitivity 25.0%).
Conclusion. The evaluated AI systems demonstrate clinically relevant performance in the emergency setting and significantly improve the diagnostic capability of emergency physicians. Their high sensitivity for fracture detection and high NPV for pulmonary nodules, pneumothorax, and fractures consolidate their role as a high-impact safety system.
