MEDDxAgent introduces a unified modular framework for automated differential diagnosis using large language models, designed to mimic physician reasoning with transparency.

• Employs an iterative approach that refines diagnoses based on symptoms and medical knowledge
• Features explainable decision-making that helps clinicians understand model reasoning
• Addresses key limitations of existing approaches through modular components that can be independently optimized
• Provides more realistic patient interactions compared to single-attempt diagnosis systems

This research represents a significant advancement in clinical decision support tools by creating AI systems that can work alongside physicians with transparent reasoning and adaptable frameworks for real-world diagnostic scenarios.

MEDDxAgent: A Unified Modular Agent Framework for Explainable Automatic Differential Diagnosis