Search
A recent study featured in Nature Medicine has brought to light a significant concern: ChatGPT Health, a medical chatbot crafted by OpenAI, frequently falls short in accurately assessing the gravity of medical emergencies during the critical case triage process. To conduct their evaluation, the research team fed 60 authentic medical cases into the system and juxtaposed its triage suggestions against the expert opinions of three seasoned clinicians. Their findings unveiled a troubling trend: ChatGPT Health exhibited a notably higher error rate when dealing with cases of extreme urgency. For instance, it often misclassified situations demanding immediate medical intervention as merely requiring attention within a 24-48 hour window, thereby heightening the risk of overlooking life-threatening emergencies.
-
C114 Communication Network -
Communication Home
