4128 Evaluating the impact of artificial intelligence-assisted image analysis on the diagnostic accuracy of front-line clinicians in detecting fractures on plain X-rays (FRACT-AI)
Rattachement africain : gb. Niveau de preuve : code pays fourni par la source.
Le résumé fourni par la source
Aims and Objectives Incorrect fracture diagnosis is the most frequent diagnostic error in UK emergency departments, causing significant patient morbidity and cost to the NHS. This study aimed to test the impact of an artificial intelligence (AI)-assisted fracture detection tool on NHS clinicians’ diagnostic performance. Method and Design A dataset of 500 plain radiographs from Oxford University Hospitals was curated based on the frequency of fracture detection claims reported by NHS Resolution, with 250 images containing one or more fractures and 250 without. Ground truth was established through independent reporting by two senior musculoskeletal radiologists, with a third arbitrating disagreement. Images were subsequently inferenced by fracture detection software (Gleamer BoneView). A multicase multireader study was conducted including 18 clinicians of varying seniority from six clinical specialties. Readers interpreted all images, then repeated the reads after a four-week washout period with AI assistance. Changes in diagnostic performance, confidence, and reporting speed were compared, along with the diagnostic performance of the algorithm against ground truth. Results and Conclusion Pooled analyses for per-case reader performance demonstrated an increase in the area under the receiver operating curve from 0.883 (95% CI 0.858–0.907) without AI to 0.921 (95% CI 0.901–0.942) with AI (p < 0.001). Sensitivity improved from 0.828 (95% CI 0.788–0.868) to 0.867 (95% CI 0.827–0.906) with AI (p < 0.001), and specificity increased from 0.829 (95% CI 0.784–0.875) to 0.904 (95% CI 0.877–0.931) with AI (p < 0.001).AI-assisted fracture detection showed potential to improve clinician and radiologist accuracy on plain X-rays. Further real-world prospective studies are needed to demonstrate clinical efficacy in urgent care settings.
Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.
Le contrôle bibliographique ouvert
DOI retrouvé dans Crossref DOI retrouvé ; titre concordant.
- Titre Crossref
- 4128 Evaluating the impact of artificial intelligence-assisted image analysis on the diagnostic accuracy of front-line clinicians in detecting fractures on plain X-rays (FRACT-AI)
- Date Crossref
- 01/04/2026
- Éditeur
- BMJ Publishing Group Ltd and the British Association for Accident & Emergency Medicine
- Type
- proceedings-article
Ce recoupement confirme des métadonnées liées au DOI. Il ne confirme ni la méthode ni les conclusions de l’étude, et il ne compte pas comme une seconde source scientifique indépendante.
Où se fait cette recherche
-
John Radcliffe Hospital pays non établi dans la noticeÉtablissement de santé
-
Oxford University Hospitals NHS Trust pays non établi dans la noticeÉtablissement de santé
-
University College London Hospitals NHS Foundation Trust pays non établi dans la noticeÉtablissement de santé
-
University College London pays non établi dans la noticeUniversité ou école supérieure
-
Great Ormond Street Hospital pays non établi dans la noticeÉtablissement de santé
-
University of Liverpool pays non établi dans la noticeUniversité ou école supérieure
-
Aintree University Hospitals NHS Foundation Trust pays non établi dans la noticeÉtablissement de santé
-
Liverpool University Hospitals NHS Foundation Trust pays non établi dans la noticeÉtablissement de santé
-
Oxford University Hospitals NHS Foundation Trust pays non établi dans la noticeUniversité ou école supérieure
-
Grest Ormond Street Hospital for Children pays non établi dans la noticeÉtablissement de santé
John Radcliffe Hospital, Oxford University Hospitals NHS Trust et University College London Hospitals NHS Foundation Trust, avec 7 autres affiliations.
Une affiliation ne permet pas de déduire la nationalité d’un auteur.