The use of Artificial Intelligence (AI), specifically GPT-4, a language model developed by OpenAI, is being explored in the field of emergency medicine to help triage patients and predict hospital admissions.
In a study published in JAMA Network Open, researchers from the University of California, San Francisco, tested GPT-4’s ability to identify patients with more severe conditions. The AI was accurate in 89% of cases, slightly surpassing the accuracy of physicians in a subgroup of 500 pairs of patient information. This suggests that AI could assist physicians in making efficient decisions, particularly in scenarios where resources are limited. However, the study’s author, Christopher Williams, M.D., emphasized the need for further validation and clinical trials before its implementation in emergency departments, with a focus on eliminating racial and gender biases.
In another study published in the Journal of the American Medical Informatics Association, researchers from the Icahn School of Medicine at Mount Sinai found that GPT-4 could predict hospital admissions with 77.5% accuracy on its first try. This accuracy improved to 83% when the program was given additional data to learn from. This technology could potentially shorten patient wait times and help determine bed requirements in hospitals. However, the researchers also highlighted the need for enhancements to ensure the AI system augments, rather than complicates, the decision-making process.
While these AI tools show promise, it’s crucial to note that they simply mimic reasoning by predicting relationships between words, as stated by Ajeet Singh, M.D., a hospitalist and informaticist at Rush University Medical Center. This means that while AI can assist in the decision-making process, doctors will always need to perform their own independent evaluations to determine a patient’s treatment.
Journalists could find interesting stories by following these studies and interviewing emergency medicine experts involved in the development and trialing of such AI technologies. This could provide insights into the potential benefits and challenges of integrating AI into emergency department settings.
