Artificial Intelligence in Radiographic Fracture Diagnosis: Current Applications, Clinical Utility, and Limitations: A Narrative Review

Artificial intelligence (AI) is increasingly investigated as a supportive tool in radiographic fracture diagnosis, particularly in emergency and musculoskeletal imaging. This narrative review aims to summarize current evidence on the use of AI in fracture detection, clinician-assisted interpretation, fracture classification, workflow optimization, and implementation challenges. A literature search was conducted in PubMed using combinations of terms related to artificial intelligence, deep learning, convolutional neural networks, fracture detection, radiography, emergency radiology, and clinical decision support. Priority was given to systematic reviews, meta-analyses, diagnostic accuracy studies, reader studies, and real-world clinical evaluations. Current evidence indicates that AI-based systems can achieve high diagnostic performance in selected fracture detection tasks, especially when applied to anatomically focused radiographic datasets. Studies assessing AI-assisted interpretation suggest that these tools may improve clinicians’ sensitivity, specificity, and overall diagnostic performance, with the greatest benefit observed among less experienced readers and non-radiologist clinicians. AI may also support fracture classification and workflow optimization, although these applications remain less extensively validated. Important limitations include restricted generalizability, reduced performance in subtle or complex fractures, limited external validation, and a shortage of prospective real-world studies. Overall, AI represents a promising but still evolving adjunct in radiographic fracture diagnosis and should be developed as a decision-support tool integrated with clinical expertise.