Assessment of the Contribution of an Artificial Intelligence Tool to Help the Diagnosis of Limb Fractures in Pediatric Emergencies
- Conditions
- Fractures, Bone
- Interventions
- Diagnostic Test: radiograph interpretation with the support of the RAYVOLVE appDiagnostic Test: radiograph interpretation without the support of the RAYVOLVE app
- Registration Number
- NCT05187585
- Lead Sponsor
- Fondation Lenval
- Brief Summary
Limb fracture is a common pathology in children. It represents the first complaint in traumatology among children in developed countries. Failure to diagnose a fracture can have severe consequences in pediatric patients with growing bones, that can lead to delayed treatment, pain and poor functional recovery.
X-ray is the first tool used by doctors to diagnose a fracture. However, the diagnosis of fracture in the emergency room can be challenging. Most images are interpreted and processed by emergency pediatricians before being reviewed by radiologists (most often the day after).
Previous studies have reported the rate of misdiagnosis in fracture by emergency physicians from 5% to 15%.
A tool to investigate in diagnosing limb fractures could be helpful for any emergency physicians exposed to this condition
- Detailed Description
Limb fracture is a common pathology in children with trauma. It represents the first complaint in traumatology among children in developed countries.
Failure to diagnose a fracture on an X-ray can have severe consequences in pediatric patients, with growing bones, that can lead to delayed treatment, pain and poor functional recovery (with risk of bone deformity and bad consolidation).
X-ray is the first tool used by doctors to diagnose a fracture. However, the diagnosis of fracture in the emergency room can be challenging. Most images are interpreted and processed by both residents and pediatricians before the radiologists proofread (most often the day after).
Previous studies have reported the rate of misdiagnosis in fracture by emergency physicians from 5 to 15%.
A tool to investigate in diagnosing limb fractures could be helpful for any clinician exposed to this condition.
Artificial intelligence (AI) in medicine is booming and has already proven its worth, in terms of prevention, monitoring and diagnosis.
AZMED has created RAYVOLVE®, a deep learning algorithm to help physicians in diagnosing fractures. The RAYVOLVE® tool connects to the PACS (Picture Archiving and Communication System) of any hospital and indicates, using a frame, the location of a potential fracture.
The tool has not yet been validated in pediatric patients.
The purpose of this research project is to evaluate the contribution of this artificial intelligence-based tool in the diagnosis of limb fracture in pediatric population.
The investigators will study the concordance in diagnosing limb fracture between the junior emergency physicians using the RAYVOLVE® application and senior radiologists, as the gold standard.
Recruitment & Eligibility
- Status
- COMPLETED
- Sex
- All
- Target Recruitment
- 1200
- Children under 18
- Showing signs that may suggest a limb fracture and justifying the realization of an X-ray (trauma with pain, deformation, edema, wound)
- Written informed consent from one of the two parents or the holder of parental authority signed
- Beneficiaries or members of a Health Insurance scheme
- A sign (s) of vital distress
- Any other reason than that of a suspected limb fracture
- A diagnosis of a limb fracture before its management in the emergency room (x-ray made in pre-hospital)
Study & Design
- Study Type
- INTERVENTIONAL
- Study Design
- PARALLEL
- Arm && Interventions
Group Intervention Description radiograph interpretation with the support of the RAYVOLVE app radiograph interpretation with the support of the RAYVOLVE app - radiograph interpretation without the support of the RAYVOLVE app radiograph interpretation without the support of the RAYVOLVE app -
- Primary Outcome Measures
Name Time Method Diagnosis fracture with Rayvolve app compare to gold standard at inclusion Assess the statistical concordance between residents using the RAYVOLVE application tool and senior radiologists in diagnosing fractures of the extremities, as gold standard.
Criteria: binary: fracture Yes/No
- Secondary Outcome Measures
Name Time Method satisfaction of the residents using the application assessed by Likert scale through study completion, an average of 6 months measure of satisfaction by an in-house Likert scale: consisting of 4 questions with multiple choice answers on the use and ergonomics of the application. The answers range from not at all satisfied to very satisfied.
Diagnosis fracture with Rayvolve app compare to diagnosis done by physicians at inclusion Assess the statistical concordance between residents using the RAYVOLVE application tool and pediatric emergency physicians in diagnosing fractures of the extremities Criteria: binary: fracture Yes/No
collection of patient data to define risk factors associated with the discrepancy between residents using the RAYVOLVE application tool and senior radiologists not using the application at inclusion collection patient data such as patient's age, fracture location, fracture type, number of fractures, day and time of diagnosis. The goal is to define potential risk factors to explain diagnostic differences between residents and primary radiologists
Diagnosis fracture without Rayvolve app compare to diagnosis done by physicians at inclusion Assess the statistical concordance between residents not using the RAYVOLVE application tool and pediatric emergency physicians in diagnosing fractures of the extremities Criteria: binary: presence or no fracture
Trial Locations
- Locations (1)
Hopitaux Pediatriques de Nice Chu-Lenval
🇫🇷Nice, France