Background: Most hip fractures occur among elderly people. They are usually treated in the emergency room where orthopedic surgeons may not be readily available. The problem of delayed diagnosis and treatment results increase risks of further complications and mortality rate. Thus, applying artificial intelligence (AI) can assist physicians having limited experience to rapidly and confidently diagnose hip fractures using radiographs. Objective: This study aimed to validate AI programs to assist diagnosing of hip fractures on plain radiographs. Methods: This study employed a retrospective diagnostic study design. From 1 January 2015 to 31 December 2019, compiled ortho pelvis, anterior-posterior (AP) films from the diagnosis of hip fractures at Ananthamahidol Hospital were performed. The performance of the AI program was compared with one orthopedic surgeon who reviewed the same images. The accuracy, sensitivity and specificity of the diagnosis of hip fractures between the orthopedic surgeon and AI program were analyzed. Results: In total, 217 patients were enrolled in this study. Of these, 56 (28.5%) were male and 161 (74.2%) female. Areas of hip fractures were as follow: intertrochanteric (108, 49.8%), femoral neck (102, 47.0%), subtrochanteric (6, 2.7%) and femoral head (1, 0.5%). The orthopedic surgeon and AI program revealed an accuracy of 93.59% (95%CI 90.8-95.73) vs. 81.24% (95% CI 77.17-84.85), sensitivity of 90.30% (95% CI 85.60-93.90) vs. 89.40% (95%CI 84.50-93.20) and specificity of 97.10% (95%CI 93.60-98.90) vs. 72.5% (95%CI 65.90-78.50), respectively. Conclusion: Our results showed that the AI model (VGG16) showed a sensitivity of 89.40% vs. 90.30% obtained from the orthopedic surgeon. Thus, improvement in the sensitivity and specificity of AI software is further required. In the future, AI models have the potential as useful tools for emergent screening and evaluation of patients with hip fractures using plain radiographs, especially in the Emergency Department where orthopedic surgeons may not be readily available.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.