Using an internal chain of debate pattern to structure multi-agent reasoning flows for diagnostic accuracy in medical AI.
