Comprehensive Summary
This systematic review of 29 studies evaluates how AI tools are currently used to detect rib fractures on X-ray and CT scans, and compares physician performance to AI. Studies were identified from two databases and were included if they used a deep-learning model to identify rib fractures on X-rays or CT scans. The review found that the majority of current studies compared AI to physicians on the same set of images for fracture detection. Radiologists were the most commonly evaluated physicians, with one study focusing on orthopedic surgeons and another on anesthesiologists. The review showed that from all studies, AI had a pooled sensitivity of 0.861 (range = 0.411–0.967), while physicians had a pooled sensitivity of 0.750 (range = 0.258–0.970). This shows that on average, AI outperformed physicians in the accurate detection of rib fractures from diagnostic images. Overall, findings show that AI algorithms are capable of accurately detecting rib fractures and have potential to improve this area in clinical practice.
Outcomes and Implications
AI-assisted detection of rib fractures has potential to significantly improve clinical efficiency and diagnostic accuracy. By reducing the likelihood of missed diagnoses on initial image review, AI-based tools can support earlier diagnosis, faster treatments, and better outcomes. Additionally, these tools can reduce the time it takes to read images, potentially saving physicians time and allowing them to address more cases in the same timeframe. Reducing diagnostic errors can also minimize unnecessary costs for healthcare systems and patients. Importantly, this review suggests that AI should serve as an assistive tool for physicians, rather than a replacement. As AI tools continue to expand throughout orthopedics and other medical specialties, their role in clinical settings is expected to grow.