Freely accessible large language models for parent education in pediatric immune thrombocytopenia: an expert-rated cross-sectional study of safety, readability, and guideline concordance

1

The study evaluated the quality and safety of responses from large language models regarding pediatric immune thrombocytopenia (ITP) for parent education.
2

Three models, GPT-5.3-mini, Gemini 3 Flash, and Claude Sonnet 4.6, were assessed using a seven-domain scoring system for their responses to ITP questions.
3

Gemini 3 Flash achieved the highest mean score for communication quality, while safety and guideline concordance were similar across all models.
4

Two instances of unsafe content were identified in Gemini 3 Flash responses, highlighting potential risks in AI-generated health information.
5

The study emphasizes the importance of clear, safe, and guideline-aligned information for parents seeking knowledge about childhood ITP.

1