Optimization of the Diagnosis of Bone Fractures in Patients Treated in the Emergency Department by Using Artificial Intelligence for Reading Radiological Images in Comparison With Traditional Reading by the Emergency Doctor.

As part of the management of a patient with suspected bone fractures, emergency physicians are required to make treatment decisions before obtaining the imaging reading report from the radiologist, who is generally not available only a few hours after the patient's admission, or even the following day. This situation of the emergency doctor, alone interpreting the radiological image, in a context of limited time due to the large flow of patients to be treated, leads to a significant risk of interpretation error. Unrecognized fractures represent one of the main causes of diagnostic errors in emergency departments.

This comparative study consists of two cohorts of patients referred to the emergency department for suspected bone fracture. The first will be of interest to patients whose radiological images will be interpreted by the reading of the emergency doctor systematically doubled by the reading of the artificial intelligence. The other will interest a group of patients cared for by the simple reading of the emergency doctor.

All of the images from both groups of patients will be re-read by the establishment's group of radiologists no later than 24 hours following the patient's treatment.

A centralized review will be provided by two expert radiologists. Also, patients in both groups will be systematically recalled in the event of detection of an unknown fracture for hospitalization.