Accurate Emergency Department (ED) diagnosis can literally be a matter of life and death. Initial research explored the use of ChatGPT in emergency treatment settings.
Chat generative pre-trained transformer (ChatGPT) continues to be tried and tested in settings and for services throughout industry. In healthcare, as we have discussed, the technology is making steps in a variety of medical fields including radiology and primary care. A small new study published in the Annals of Emergency Medicine suggests an AI-trained Chatbot shows promise in providing Emergency Department diagnosis.
While trained for the task, physicians in the ED face a daunting job. ED doctors must effectively and accurately triage patients coming in the door, attend to or restore their physical stability, and provide effective guidance on next steps. Unlike a primary care physician, ED physicians usually do not have the advantage of a pre-existing relationship. As well, many EDs are over-crowded, fast paced, and potentially understaffed.
Like the rest of the medical industry, emergency medicine continues to evolve. Researchers in this study were interested in understanding if AI can be of assistance in providing diagnostic services.
To compare how a Chatbot might perform when diagnosing some of these conditions, the study authors looked at data on 30 patients who were treated at hospital in the Netherlands in 2022. Physician notes, symptoms, lab results, and exam results were uploaded to the two existing version of ChatGPT (3.5 and 4.0). Researchers then compared the short list of diagnoses created by a human physician and the Chatbot.
The study found 60 percent of the possible diagnoses matched between human and bot. In an interesting turn, human providers listed the correct diagnosis within their list in 87 percent of cases. The bot listed the correct diagnosis 97 percent of the time for version 3.5 and 87 percent of the time for version 4.0.
Study author Dr. Hidde ten Berg, said “We found that ChatGPT performed well in generating a list of likely diagnoses and suggesting the most likely option. We also found a lot of overlap with the doctors’ lists of likely diagnoses. Simply put, this indicates that ChatGPT was able to suggest medical diagnoses much like a human doctor would.”
Researchers stress more studies are needed to identify the true potential of AI in this and other medical settings. But the findings of this study are compelling—two versions of ChatGPT diagnostically equaled and potentially out-performed experienced ED physicians. And—like anywhere else in healthcare, ChatGPT is as likely to make an injurious or fatal medical mistake along the way. Despite the booming adoption of AI, it may be some time before ChatGPT is used—reliably and responsibly—in an ED.
Reach out to our law firm if you or a loved one is injured by medical malpractice
Schochor, Staton, Goldberg and Cardea, P.A. has been fighting for patients’ rights as a leading medical malpractice law firm since 1984. The firm is a recognized leader in medical malpractice, sexual abuse matters, and class action and mass tort litigation. Contact us or call 410-234-1000 to schedule a free consultation. We have offices in Washington, DC and Baltimore, Maryland.