PediaBench introduces a specialized dataset for evaluating large language models' performance in pediatric healthcare scenarios.

• Covers 12 pediatric disease groups with both objective questions and clinical case analyses
• Designed to assess both medical knowledge and practical clinical reasoning abilities
• Creates a standardized evaluation framework specifically for the underrepresented pediatric domain

This research addresses a critical gap in medical AI benchmarking by focusing on pediatric-specific knowledge, enabling more accurate assessment of LLMs for potential deployment in children's healthcare settings.

PediaBench: A Comprehensive Chinese Pediatric Dataset for Benchmarking Large Language Models