Assessing Artificial Intelligence-Generated Patient Educational Material on Gestational Diabetes Mellitus: Content and Quality Evaluation.

PURPOSE: This study aims to evaluate the content and quality of patient educational materials on gestational diabetes mellitus (GDM) generated by ChatGPT and Gemini.

BACKGROUND

The sources of knowledge are crucial in the effective management of disease. Artificial intelligence (AI) platforms could become a primary source of patient education materials in the near future.

METHODS

A descriptive research design was employed. Frequently asked questions related to GDM were extracted from patient education sections of existing guidelines. These questions were then submitted to both ChatGPT and Gemini. The responses provided by these platforms were used to create educational material aimed at pregnant women diagnosed with GDM. The content was reviewed by a panel of 11 experts. The Patient Education Materials Assessment Tool for Printed Materials (PEMAT-P) was employed to evaluate the content's effectiveness and clarity, and the readability was assessed through the Ateşman Readability Formula and the Gunning Fog Index.

RESULTS

A total of 32 questions regarding GDM were directed to the AI platforms. The resulting educational materials had a readability score of 77.8 based on the Ateşman scale and 16.25 according to the Gunning Fog Index. The experts rated the material as highly comprehensible, with an average PEMAT-P understandability score of 91.36% (range: 86.66%-93.75%) and an actionability score of 89.67% (range: 80%-100%).

CONCLUSION

The GDM educational materials generated by ChatGPT and Gemini exhibit a high level of readability, making them easy to understand. Moreover, the material was deemed comprehensible and actionable for pregnant women with GDM.

IMPLICATIONS FOR PRACTICE AND RESEARCH: Although AI-generated patient educational materials show great potential, further experimental research is necessary to assess their long-term effectiveness.