Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Jul 16, 2026

Imaging of the Microstructural Failure Mechanism in the Human Hip
08:43

Imaging of the Microstructural Failure Mechanism in the Human Hip

Published on: September 29, 2023

Between Accessibility and Reliability: High Confidence, Low Control in General-Purpose Multimodal Models for Hip

Hadar Gan-Or1, Shaked Ankol2, Guy Ben Arie3

  • 1Department of Orthopedic Surgery, Hillel Yaffe Medical Center, Hadera 3820302, Israel.

Journal of Clinical Medicine
|July 15, 2026
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Factors associated with substantial preoperative haemaglobin decline in hip fracture patients: a propensity-matched analysis of 1935 cases.

Hip international : the journal of clinical and experimental research on hip pathology and therapy·2026
Same author

Evaluation of Experience, Training, and Hand Dominance on Drilling Accuracy in Orthopedic Surgeons-A Preliminary Study.

Medicina (Kaunas, Lithuania)·2026
Same author

The Hidden Risks of Hip Replacement: Unveiling Mortality and Costs in 1.6 Million Patients.

Healthcare (Basel, Switzerland)·2025
Same author

The Impact of Concomitant Upper Extremity Fractures on Outcomes in Geriatric Patients Following Hip Fracture Surgery.

Journal of clinical medicine·2025
Same author

Comparative Analysis of Cervical Disc Arthroplasty and Anterior Cervical Discectomy and Fusion: Trends, Demographics, and Clinical Outcomes in a Nationwide Inpatient Sample.

Journal of clinical medicine·2025
Same author

Does a Dual-Mobility Cup Offer Better Stability than Conventional Bearings in Hip Arthroplasty Following Femoral Neck Fracture?

Journal of clinical medicine·2025

General-purpose AI models show limited accuracy in detecting hip fractures, often remaining overconfident when incorrect. These AI tools are not yet ready for unsupervised clinical use in orthopedic imaging.

Area of Science:

  • Artificial Intelligence in Medicine
  • Orthopedic Imaging Analysis
  • Medical Device Validation

Background:

  • Dedicated AI for fracture detection exists, but general-purpose multimodal models are increasingly accessible to clinicians.
  • These general models are not validated medical devices, and their performance in orthopedic tasks is undercharacterized.
  • The study focuses on the interpretation of AP pelvis radiographs with hip fractures by these accessible AI models.

Purpose of the Study:

  • To characterize the interpretation of AP pelvis radiographs with hip fractures by two general-purpose multimodal AI models.
  • To evaluate context dependence, overconfidence, and error patterns in AI models within a fracture-positive cohort.
  • To assess AI performance on fracture detection, laterality, fracture patterns, and AO/OTA grading.
Keywords:
AI safetyAI-assisted radiograph interpretationfracture classificationhip fracturemodel confidencemultimodal large language modelspelvic radiographyprompt sensitivity

More Related Videos

The Use of Mixed Reality in Custom-Made Revision Hip Arthroplasty: A First Case Report
07:45

The Use of Mixed Reality in Custom-Made Revision Hip Arthroplasty: A First Case Report

Published on: August 4, 2022

Related Experiment Videos

Last Updated: Jul 16, 2026

Imaging of the Microstructural Failure Mechanism in the Human Hip
08:43

Imaging of the Microstructural Failure Mechanism in the Human Hip

Published on: September 29, 2023

The Use of Mixed Reality in Custom-Made Revision Hip Arthroplasty: A First Case Report
07:45

The Use of Mixed Reality in Custom-Made Revision Hip Arthroplasty: A First Case Report

Published on: August 4, 2022

Main Methods:

  • Retrospective study of 214 surgically confirmed hip fractures on AP pelvis radiographs.
  • Utilized two general-purpose multimodal AI models under six different prompting conditions.
  • Recorded fracture detection, laterality, fracture pattern, displacement, AO/OTA grading, and confidence levels.

Main Results:

  • Combined AI models achieved 89.7% and 96.7% accuracy for hip fracture detection with correct laterality under different conditions.
  • Incorrect laterality cues significantly degraded AI performance.
  • AI models showed limited performance in treatment-relevant subtyping (AO/OTA grading) and remained highly confident when incorrect.

Conclusions:

  • General-purpose multimodal AI models demonstrate partial capability for coarse hip fracture interpretation but are context-sensitive and unreliable for detailed subtyping.
  • High confidence in incorrect predictions highlights limitations for clinical readiness.
  • The study advises against unvalidated clinical use, emphasizing the need for usage boundaries, revalidation, and human oversight.