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

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

  • By

  • Orkun Dinç

  • Eray Akay

  • Belen Ates

  • July 1, 2026

  • 0 min

Share

Clinical Report: Evaluation of Open-Access Large Language Models for Educating Parents on Pediatric Immune Thrombocytopenia

Overview

This study evaluates the quality, safety, readability, and guideline concordance of responses from large language models (LLMs) to parent-oriented questions about pediatric immune thrombocytopenia (ITP).

Background

Pediatric immune thrombocytopenia (ITP) is a common acquired bleeding disorder in children, characterized by low platelet counts. Effective communication of information regarding ITP is crucial for parents. The increasing reliance on online resources for medical information necessitates an evaluation of the quality and safety of these resources, particularly those generated by LLMs.

Data Highlights

ModelMean Composite Score (AI-ITP-PRS)Standard Deviation
Gemini 3 Flash32.782.54
Claude Sonnet 4.630.601.39
GPT-5.3-mini29.981.52

Key Findings

  • Significant differences in the AI-ITP-PRS scores across models (Friedman chi-square = 63.52, p < 0.001).
  • Gemini 3 Flash had the highest mean composite score, driven by completeness and comprehensibility.
  • Medical accuracy and guideline concordance were largely similar across models.
  • Two instances of unsafe content were identified in Gemini 3 Flash responses.
  • Readability was assessed using automated indices, indicating varying levels of comprehensibility.

Clinical Implications

Healthcare providers should be aware of the variability in the quality of information provided by LLMs when educating parents about pediatric ITP.

Conclusion

The study identifies areas for improvement in safety and accuracy of LLMs in providing educational content for parents of children with ITP.

Related Resources & Content

  1. JMIR Medical Informatics, 2026 -- Evaluation of Five Large Language Models for Parental Education in Pediatric Anesthesia: Reliability and Readability Study
  2. Frontiers in Psychiatry, 2026 -- Assessing Large Language Model Responses to Pediatric Depression FAQs: A Cross-sectional Study on Readability, Accuracy, and Sentiment
  3. Frontiers in Pediatrics, 2026 -- Development and validation of a predictive model for chronic or persistent immune thrombocytopenia in children incorporating anti-glycoprotein IIb antibody
  4. Immune Thrombocytopenia | Pediatric Care Online | American Academy of Pediatrics
  5. Eltrombopag for Newly Diagnosed Pediatric Immune Thrombocytopenia Requiring Treatment: The PINES Randomized Clinical Trial - PubMed
  6. Drugs - Real World Outcomes — Evaluating Pediatric Medication Safety and Effectiveness Through Real-World Evidence: A Systematic Review
  7. Immune Thrombocytopenia | Pediatric Care Online | American Academy of Pediatrics
  8. Eltrombopag for Newly Diagnosed Pediatric Immune Thrombocytopenia Requiring Treatment: The PINES Randomized Clinical Trial - PubMed
  9. Working toward international consensus defining pediatric “refractory ITP” - PMC

Original Source(s)

Related Content