Related Experiment Video
Updated: Mar 27, 2026
Orthopedic Robot-Assisted Femoral Neck System in the Treatment of Femoral Neck Fracture
Published on: March 3, 2023
Artificial intelligence-assisted triage of pediatric supracondylar humerus fractures in emergency departments: A
Neeraj Mishra1, Nicole Kim Luan Lee2, Andrew Chia Chen Chou1
1Department of Orthopaedic Surgery, KK Women's and Children's Hospital, Singapore.
Insights
An artificial intelligence (AI) model shows promise in classifying pediatric supracondylar humerus fractures (SCHF) and aiding treatment decisions. This AI tool achieved diagnostic consistency comparable to human experts, supporting emergency department triage.
Area of Science:
- Orthopaedic Surgery
- Medical Imaging
- Artificial Intelligence
Background:
- Pediatric supracondylar humerus fractures (SCHF) are common pediatric elbow injuries.
- Accurate Gartland classification is crucial for treatment, but inter-observer variability is significant, especially among non-specialists.
- Artificial intelligence (AI) offers potential for decision support in emergency settings with limited specialist access.
Purpose of the Study:
- To develop and validate an AI model for classifying pediatric SCHF.
- To assess the AI model's ability to triage fractures into operative versus non-operative management categories.
Main Methods:
- Retrospective analysis of 1811 annotated pediatric SCHF radiographs (2010-2017).
- Fractures dichotomized into operative (Gartland IIB, III, flexion) and non-operative (normal, I, IIA) groups.
- A convolutional neural network was trained and tested on independent radiograph sets, with performance evaluated using accuracy, sensitivity, specificity, and Cohen's kappa.
Main Results:
- The AI model achieved 77% overall accuracy and an AUROC of 0.84.
- Specificity was 82% and sensitivity was 69%.
- Inter-observer agreement between the AI and surgeon consensus (κ=0.51) was comparable to individual clinician variability. Grad-CAM localized predictions to the supracondylar region in 83% of cases.
Conclusions:
- AI can serve as a feasible decision-support tool in emergency departments for pediatric SCHF classification.
- The model demonstrated diagnostic consistency within the range of human inter-observer variability.
- Further multi-center validation and safety protocols are recommended before widespread implementation.
Background:
Pediatric supracondylar humerus fractures (SCHF) are the most common elbow fractures in children. Accurate classification using the Gartland system is essential for treatment decisions, yet considerable inter-observer variability persists, particularly among non-specialist emergency physicians. Artificial intelligence (AI) has potential to support triage in emergency settings where specialist orthopaedic expertise may be limited. The aim of this study was to develop and validate an AI model capable of assisting in the classification of paediatric SCHF and triaging cases into operative versus non-operative management categories.
Methods:
We retrospectively analysed 1811 annotated radiographs (anteroposterior and lateral views) of paediatric SCHF from a single tertiary Paediatric Orthopaedic centre (2010-2017). Fractures were dichotomised into operative (Gartland IIB, III, flexion type) and non-operative (normal, I, IIA). Two senior Paediatric Orthopaedic surgeons performed independent labelling, with consensus resolution of discordance. A convolutional neural network model was trained with transfer learning and tested on an independent set of 215 radiograph pairs. Test-set composition: 128 non-operative (59.5 %), 87 operative (40.5 %). Model performance was assessed using accuracy, sensitivity, specificity, predictive values, AUROC, and Cohen's kappa.
Results:
The model achieved overall accuracy of 77 % (95 % CI 71-82 %) and AUROC of 0.84 (95 % CI 0.78-0.89). Specificity (82 %, 95 % CI 74-88 %) exceeded sensitivity (69 %, 95 % CI 58-78 %). Inter-observer agreement between the AI model and surgeon consensus was κ = 0.51 (95 % CI 0.39-0.63), reflecting diagnostic consistency comparable to individual clinician interpretation. Grad-CAM visualisations localised to the supracondylar region in 83 % of predictions.
Conclusions:
This validation study demonstrates that AI can feasibly assist as a decision-support filter in emergency departments, achieving diagnostic consistency within the range of documented human inter-observer variability. All high-probability predictions require mandatory surgeon review before operative decisions. This proof-of-concept supports further development through multi-centre external validation, expanded governance frameworks, and deployment safety protocols before implementation in diverse emergency settings.
Level Of Evidence:
II.
Related Concept Videos
01:27Fractures: Bone Repair
Minor fractures with no bone displacement are treated by immobilizing the fractured bone using a cast or splint. However, in the case of fractures with displaced bones, the broken bones are repositioned before immobilization to ensure successful healing without deformation and loss of function. The realignment of fractured bone ends is performed through a process called reduction. If the...
01:26Flail Chest-II
Assessment:
1. Clinical Evaluation:
History: