Comparative analysis reveals ChatGPT's diagnostic accuracy for gastrointestinal tract perforation in chest imaging, indicating potential enhancements over human evaluation.
Key Points
ChatGPT demonstrated comparable diagnostic accuracy to human experts in identifying gastrointestinal tract perforation.
The study analyzed 504 chest radiographs, with classifications into definite perforation, suspicious cases, and no perforation.
Evaluation included assessments from emergency medicine specialists alongside ChatGPT models 3.5 and 4.
Results highlight the potential for artificial intelligence to improve diagnostic performance in emergency medicine.