The role of artificial intelligence in gynecologic and obstetric emergencies

To investigate the potential of artificial intelligence (AI) in emergency medicine, focusing on its utility in triaging and managing acute gynecologic and obstetric emergencies.

This feasibility study assessed Chat-GPT's performance in triaging and recommending management interventions for gynecologic and obstetric emergencies, using ten fictive cases. Five common conditions were modeled for each specialty. Chat-GPT was tasked with proposing triage classifications and providing immediate management recommendations. Human experts independently reviewed each case, classified triage categories, and proposed management. Following this, experts evaluated Chat-GPT's recommendations, rating the AI's responses on accuracy and clinical applicability.

Chat-GPT's recommendations demonstrated high concordance with human evaluators. Chat-GPT's triage classifications matched those of human experts in most cases, though minor discrepancies in urgency ratings were observed. The AÍs suggestions were mostly rated as "very good" to "excellent." While Chat-GPT consistently delivered appropriate responses, some human evaluators noted slight differences in perceived urgency.

This study highlights Chat-GPT's potential as a clinical support tool in emergency medicine. Chat-GPT provided structured, evidence-based recommendations comparable to those of experienced clinicians, especially for high-stakes gynecologic and obstetric emergencies. Although encouraging, these results highlight the value of utilizing AI in addition to human knowledge, as variations in urgency ratings and management nuances highlight the necessity of human supervision in crucial decision-making.