Attribute Structuring Improves LLM-Based Evaluation of Clinical Text Summaries