Accuracy, Readability, and Clinical Applicability of AI-Generated Breastfeeding Education Brochures: A Mixed-Methods...

Generative artificial intelligence (AI) is increasingly used in health communication and nursing education; however, its clinical reliability remains uncertain. Breastfeeding education requires accurate, readable, and clinically applicable materials.

The purpose of this explanatory sequential mixed-methods study was to evaluate AI-generated breastfeeding education brochures in terms of accuracy, readability, suitability, and clinical applicability as assessed by women's health nurses and midwives.

A total of 26 women's health nurses and midwives (13 per group) were recruited using convenience sampling from obstetric and neonatal units. Participants independently evaluated AI-generated brochures using a researcher-developed 20-item instrument scored on a 3-point scale (0-2). Content validity was established through expert review by five academic specialists in women's health nursing. Internal consistency reliability was high (α = .94 and α = .72). Domain comparisons were conducted using Mann-Whitney U tests for independent groups. Semi-structured interviews (approximately 20 min) were thematically analyzed to contextualize quantitative findings.

Readability received the highest mean scores (M = 1.68-1.71), whereas clinical applicability yielded the lowest scores (M = 1.18-1.44). Although slight differences in mean scores were observed across domains, no statistically significant differences were found between groups (p > 0.05). Qualitative findings indicated strengths in clarity and structure but highlighted insufficient safety warnings and lack of source citation.

AI-generated breastfeeding brochures demonstrate strong readability but require expert oversight to ensure clinical safety and applicability. These findings support the integration of AI tools with professional validation in women's health education.