EMPOWER: Evolutionary Medical Prompt Optimization With Reinforcement Learning
It addresses critical challenges in developing clinically appropriate prompts for healthcare settings, facilitating more responsible integration of LLMs.
The paper tackled the problem of optimizing prompts for Large Language Models in medical applications by introducing EMPOWER, an evolutionary framework that reduced factually incorrect content by 24.7%, enhanced domain specificity by 19.6%, and increased clinician preference by 15.3%.
Prompt engineering significantly influences the reliability and clinical utility of Large Language Models (LLMs) in medical applications. Current optimization approaches inadequately address domain-specific medical knowledge and safety requirements. This paper introduces EMPOWER, a novel evolutionary framework that enhances medical prompt quality through specialized representation learning, multi-dimensional evaluation, and structure-preserving algorithms. Our methodology incorporates: (1) a medical terminology attention mechanism, (2) a comprehensive assessment architecture evaluating clarity, specificity, clinical relevance, and factual accuracy, (3) a component-level evolutionary algorithm preserving clinical reasoning integrity, and (4) a semantic verification module ensuring adherence to medical knowledge. Evaluation across diagnostic, therapeutic, and educational tasks demonstrates significant improvements: 24.7% reduction in factually incorrect content, 19.6% enhancement in domain specificity, and 15.3% higher clinician preference in blinded evaluations. The framework addresses critical challenges in developing clinically appropriate prompts, facilitating more responsible integration of LLMs into healthcare settings.