Real-Time Artificial Intelligence Diagnostic Copilot in Simulated Primary Care Consultations: Randomized Simulation Study.
Ivan Cusacovich, Manuel Pinilla, Ruben Garcia Castro
✦ AI-curated · Sources linked
In 30 seconds
This randomized simulation study evaluated the impact of a real-time, voice-based AI diagnostic assistant on physician diagnostic accuracy in primary care consultations. Involving 13 board-certified physicians managing 260 simulated cases, the AI-assisted condition achieved a Top-3 diagnostic accuracy of 74.6% compared to 62.3% in the unassisted condition, with an adjusted odds ratio of 2.68 (P = .01).
Key findings
- AI-assisted diagnostic accuracy was 74.6% compared to 62.3% in the unassisted condition.
- The adjusted odds ratio for improved accuracy with AI assistance was 2.68 (P = .01).
- Mean consultation time increased by 10.7% with AI assistance.
- Physicians rated the AI system highly for usefulness and satisfaction.
Why it matters
The study highlights the potential of AI diagnostic assistants to enhance diagnostic accuracy in primary care, addressing a significant source of diagnostic errors. Understanding their effectiveness and risks is crucial for future clinical implementation.