Differential Diagnosis

The purpose of this study is to investigate whether the use of generative artificial intelligence (AI) as a support tool can improve physicians' ability to establish accurate differential diagnoses in emergency care. The main questions it aims to answer are: * Is AI-assisted differential diagnosis more accurate than the differential diagnosis performed by physicians? * Does physicians' differential diagnosis change - in terms of the number of correct diagnoses - when they have access to AI-assisted differential diagnosis?

This study evaluates the diagnostic performance of a multimodal artificial intelligence (AI) system (AIMD.1) using de-identified medical images and semi-synthetic patient simulations. The study combines retrospective analysis of existing publicly available image datasets with prospective data collection from licensed clinicians who complete diagnostic evaluation tasks. In the One-Shot Vision Differential Evaluation (OSVDE) stage, clinicians review individual de-identified medical images and generate a ranked list of potential diagnoses based solely on visual features. In the Multi-Step Conversational Non-Inferiority Evaluation (MSCNE) stage, clinicians complete diagnostic assessments using semi-synthetic patient simulations derived from de-identified medical images. Clinician performance will be compared with the AI system on the same diagnostic tasks. Human participants consist solely of licensed clinicians who provide diagnostic responses. Medical images and simulated cases are study materials and are not considered study participants. No identifiable patient data are used, and the AI system is evaluated in an offline research environment and is not used for clinical decision-making or patient care.

Differentiating CD from intestinal tuberculosis (ITB) is difficult due to the low sensitivities of currently available diagnostic tests. The Asia-Pacific guideline recommends anti-tuberculous therapy (ATT) for 8-12 weeks in patients with diagnostic uncertainty due to the risk of disseminated tuberculosis if patients with ITB are misdiagnosed with CD, and are prescribed immunosuppressive therapy. However, treatment with ATT has many side effects and may delay treatment in patients with CD, and this may cause severe relapse and developing complications. Many studies found that some clinical, endoscopy, pathology, radiology, and serology findings can help to improve diagnostic accuracy in these patients. However, no single diagnostic parameter can distinguish between CD and ITB. As a result, many models were developed that include various factors and modalities, and many of those models have been reported to have high performance. However, the number of studies performed to validate those models externally was limited. Correspondingly, this study is designed to prospectively validate models that integrate more advanced parameters (e.g., IGRA, CT enterography findings) with clinical, endoscopic, or pathological findings. However, it aims mainly to evaluate the model integrating clinical, endoscopic, and serological variables since CT enterography and pathological interpretation require experienced radiologists and pathologists but they are not available in many centers.