CoMMa: Contribution-Aware Medical Multi-Agents From A Game-Theoretic Perspective
This game-theoretic AI could revolutionize how doctors diagnose complex cancer cases.
Researchers have introduced CoMMa, a new decentralized multi-agent AI framework designed for oncology decision support. Unlike typical narrative-based agents, CoMMa uses deterministic embedding and game theory to assign credit to individual specialist agents, creating interpretable and stable decision pathways. Tested on real-world multidisciplinary tumor board data, it achieved higher accuracy and more stable performance than centralized and role-based multi-agent baselines, marking a significant step toward reliable AI-assisted medical diagnosis.
Why It Matters
It provides a more stable, interpretable, and accurate AI system for critical, life-saving medical decisions in oncology.