LLM Clinical Support in the ED: Engagement Falls Despite Accurate Advice

A prospective ED study of a large language model (LLM) clinical decision support system found physician usage dropped from 68% to 30% due to workload. Physicians preferred using it for radiology consultations (OR 2.98). Expert review rated 99 of 100 sampled outputs as clinically…

• In this 1,138-patient DECIDE-AI stage 1 study, intention-to-treat analysis showed a non-significant trend toward shorter consultation cycle time (-9.4 min, P = 0.077).
• Workload-sensitive disengagement was the primary barrier, with adoption declining as shift hours increased (OR = 0.72 per shift hour, 95% CI 0.62 to 0.83).
• The study suggests sustained clinician engagement, not algorithmic accuracy, is the key barrier to effective AI deployment in emergency settings.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.