MACD: Multi-Agent Clinical Diagnosis with Self-Learned Knowledge for LLM
The discussion of MACD explores a multi-agent clinical diagnosis framework that processes thousands of real patient cases. The system utilizes specialized agents—Summarizer, Refiner, and Diagnostician—to create evolving, 'Self-Learned Knowledge.' This approach significantly outperforms established professional knowledge and is designed to be scalable and deployable on local hardware.