DeCode: Decoupling Content and Delivery for Medical QA

Researchers introduce DeCode, a training-free framework that adapts large language models to provide better contextualized medical answers by decoupling content from delivery. The system significantly improves clinical question answering performance, boosting zero-shot results from 28.4% to 49.8% on medical benchmarks.

• →DeCode is a training-free, model-agnostic framework that improves medical AI responses without requiring model retraining.
• →The system addresses a key limitation where LLMs give clinically correct but contextually inappropriate answers to patients.
• →Performance on OpenAI HealthBench improved from 28.4% to 49.8% in zero-shot scenarios.
• →The framework achieves new state-of-the-art results in clinical question answering compared to existing methods.
• →DeCode demonstrates the potential for better AI-assisted healthcare by improving patient-context alignment.