Developing AI-generated Storybooks for Patient Education

Abstract

Recent advances in Large Language Models and Text-to-Image technologies present new opportunities for developing innovative and personalized patient education tools. This study aimed to establish a practical framework for creating multimodal digital storybooks that combine narrative text, visuals, and audio narration to support patients with chronic diseases. A qualitative content development design was applied, using an iterative process guided by expert feedback to ensure clinical accuracy and cultural relevance. ChatGPT-5 (OpenAI) was utilized for prompt engineering, while Gemini 2.5 Pro (Google LLC) generated storybooks integrating text, visuals, and narration for 3 chronic conditions: heart failure, diabetes mellitus, and chronic obstructive pulmonary disease. Each storybook underwent multiple cycles of refinement to address clinical detail, readability, and visual accuracy. Findings indicated that while AI systems can generate patient education materials, general prompts were insufficient, and technical limitations, such as medically inaccurate visuals, required expert correction. Despite these challenges, the framework demonstrated that high-quality, accessible, and engaging storybooks can be produced when AI is used in collaboration with subject matter experts. This approach highlights AI's role as a supportive tool for health care professionals, enhancing their ability to deliver effective, patient-centered education that promotes understanding, engagement, and improved self-management in chronic disease care.

Ähnliche Arbeiten