Artificial Intelligence in Bone Fracture Detection: A Review of Evidence, Limitations, and Clinical Integration

Abstract

Medical imaging is rapidly being improved by artificial intelligence (AI), with deep-learning systems performing well in radiography, CT, and MRI for fracture detection, classification, and localization. This narrative review examined recent evidence spanning different types of bone fractures, alongside soft-tissue injuries relevant to orthopedic decision-making. Across multiple meta-analyses and external validations, reported sensitivities and specificities commonly range from 0.85 to 0.95, while AI also supports workflow triage and reader confidence. Persistent gaps include limited generalizability, inconsistent reference standards, spectrum bias, regulatory and ethical challenges, and implementation costs. We outline pragmatic quality considerations and emphasize prospective, multi-center trials and transparent reporting for safe clinical integration. AI should augment clinicians, improving speed, accuracy, and overall patient outcomes.