ELAXIR cards: Co-design and formative evaluation of a digital-physical educational tool to enhance AI literacy and awareness of the ethical risks of using AI in healthcare (Preprint)

Abstract

<sec> <title>BACKGROUND</title> Artificial intelligence (AI) is increasingly integrated into healthcare systems, yet many clinicians and patients lack foundational AI literacy and awareness of ethical risks. Educational tools that foster dialogue, reflection, and shared understanding of AI ethics are needed to support responsible adoption and informed decision-making in clinical contexts. </sec> <sec> <title>OBJECTIVE</title> This formative participatory study aimed to co-design and iteratively refine a hybrid digital-physical educational toolkit (ELAXIR cards) to (1) provide foundational AI knowledge for patients and healthcare professionals, (2) raise awareness of ethical risks and challenges in AI development and use, and (3) stimulate dialogue to support safer and more responsible AI integration in healthcare. </sec> <sec> <title>METHODS</title> A participatory mixed-methods approach was used to co-create and refine the ELAXIR toolkit. The development process was guided by adult learning theory, ethical AI frameworks, and Value Sensitive Design (VSD), ensuring human values were embedded into the educational content. The co-design process included a co-creation workshop with an extended research team (n=15), a patient and public involvement and engagement (PPIE) session (n=6), and two expert focus groups (n=10 and n=7). Iterative content creation involved drafting real-world medical scenarios, reviewing evidence-based AI applications, and designing discussion prompts to stimulate ethical reflection. Evaluation combined qualitative feedback with quantitative pre/post surveys assessing familiarity with AI and confidence in applying AI ethics principles. </sec> <sec> <title>RESULTS</title> Following the ELAXIR card activity, most participants reported increased confidence in understanding AI and ethical issues, describing the tool as easy to use and highly engaging. During the PPIE session (n=6), participants reported high baseline awareness of AI applications in healthcare and expressed concerns about data governance, privacy, fairness, and potential misuse. Most participants reported increased confidence in their understanding of AI in healthcare and felt the cards encouraged reflection on ethical challenges they had not previously considered. Patients valued the simplicity, variety, and realistic scenarios, although some suggested larger font size for accessibility. In the expert focus groups, 17 participants completed pre/post surveys; confidence in understanding bias increased from 59% to 94%, and confidence in applying AI ethics principles increased from 41% to 88%. Participants rated the toolkit as highly useful (88%) and reported strong potential to improve understanding, stimulate reflection, and support discussion with patients and colleagues. Qualitative feedback emphasized clarity, relevance, and accessibility, while recommending clearer wording, larger fonts, additional resources, and more interactive gameplay elements. Participants also suggested gamification elements to improve engagement and enable use in waiting rooms, community settings, and PPIE activities. </sec> <sec> <title>CONCLUSIONS</title> The ELAXIR cards are a feasible and acceptable hybrid educational tool that improves confidence and stimulates discussion about AI ethics in healthcare. The findings support further pilot testing to evaluate effectiveness across broader clinical and public settings. </sec>

Ähnliche Arbeiten