Evaluating the Precision of ChatGPT Artificial Intelligence in Emergency Differential Diagnosis

Abstract

Introduction: artificial intelligence (AI) is the study and development of intelligent machines that can carry out tasks that would typically require human intelligence. AI seeks to give machines the ability to think, problem-solve, sense their surroundings, and comprehend human speech. By enhancing and optimising processes, this technology is predicted to completely transform a number of industries. Artificial intelligence is tipped to be the next technological breakthrough that will shape our future. Objective: This study focused on evaluating the precision of ChatGPT artificial intelligence in emergency differential diagnosis. Methods: This was a comparison study, conducted from August to September 2023, evaluating the ability of both the Monica ChatGPT and the emergency medicine textbooks to provide differential diagnoses for frequently occurring complaints. Twelve symptoms common to adult patients were included in the list of chief complaints. To gauge the accuracy of the ChatGPT’s answers, the researcher employed ChatGPT®-4 queries. Results: The total number of differential diagnoses captured by the two resources was 431. The ChatGPT captured a total of 272 differential diagnoses; however, 59 of these were not included in the list of the chief complaints. Conclusion: The study concludes that AI can be helpful in some situations, such as primary care diagnosis and patient triage, although in most cases it is not a better diagnostic tool. Therefore, AI and human diagnosis can be used concurrently in the health sector.

Keywords

• Artificial Intelligence
• ChatGPT
• Differential Diagnoses
• Emergency, Evaluation
• Precision