Opracowanie serii książeczek edukacyjnych w oparciu o współpracę człowieka i sztucznej inteligencji do nauki o epilepsji i wspierania samodzielnego zarządzania chorobą
PubMed➕ 04.07.2026Epilepsy Behav
Development of a human-artificial intelligence collaboration-based storybook series for understanding epilepsy and supporting self-management
W skrócie
Badacze stworzylidziewięć cyfrowych książeczek o epilepsji przy pomocy sztucznej inteligencji, wspieranej przez ekspertów medycznych, które pomagają pacjentom zrozumieć chorobę i uczyć się, jak ją kontrolować. Materiały poruszają ważne tematy takie jak przyjmowanie leków, poznawanie czynników wyzwalających, radzenie sobie z emocjami i udzielanie pierwszej pomocy. Ocena przez specjalistów wykazała, że książeczki są dobrze zrozumiałe i przydatne dla pacjentów, choć sztuczna inteligencja czasami miała problemy z najprecyzyjniejszymi informacjami medycznymi, dlatego człowiek musi zawsze nadzorować takie rozwiązania.
Oryginalny abstract (angielski)
Epilepsy is a chronic condition that requires ongoing self-management, including medication adherence, trigger control, lifestyle regulation, and psychosocial coping. Patient education improves treatment adherence and quality of life; however, current educational materials are often text-heavy, time-consuming to produce, and limited in addressing emotional and cognitive learning needs. This study aimed to develop and expert-validate a human-AI collaborative multimodal storybook series with the potential to support epilepsy education and strengthen self-management competencies. Nine AI-assisted digital storybooks were produced using Google Gemini 2.5 Pro through iterative prompt engineering and expert-led refinement. The researchers ensured conceptual accuracy, narrative integrity, and educational alignment, and a medical text-locking protocol was applied to prevent AI-generated misinformation in high-risk areas, including medication guidance and seizure first aid. Storylines followed core self-management pathways, addressing diagnosis, seizure characteristics, medication safety, trigger awareness, stigma, emotional fluctuations, and first-aid response. Visual and narrative components were repeatedly optimized to enhance clarity, developmental relevance, and learning coherence. Five experts evaluated the materials using the DISCERN and PEMAT-A/V tools. Results showed high reliability (mean DISCERN score = 71.03 ± 1.09) and excellent understandability and applicability (PEMAT-A/V scores of 92.22 and 100). AI demonstrated strong performance in storytelling and simplification of medical concepts, while limitations were observed in nuanced clinical reasoning, visual accuracy, and interface structuring, reinforcing the need for expert oversight. Human-AI collaboration may enable rapid development of accessible, accurate, and engaging educational resources, suggesting potential as a scalable approach for digital epilepsy self-management support.